Center for Universal Education Working Paper 6 | AUGUST 2012

Center for Universal Education at BROOKINGS


Kevin Watkins Woubedle Alemayehu Center for Universal Education at BROOKINGS Kevin Watkins is a non-resident senior fellow at the Brookings’ Center for Universal Education.

Woubedle Alemayehu is a research project manager for Georgetown University.


We would like to thank the many people who helped to shape this report. Special mention must be made of Hon. Mohamed Elmi, the minister of state for Development of Northern Kenya and Other Arid Lands (MDNKOAL), who read and commented on an early draft. Other staff in the ministry also helped to guide our work. Izzy Birch, technical advisor to the minister, provided a mixture of patient guidance, support and encouragement. While we spare Izzy any responsibility for the final content of the report, her insights have been invaluable—and we owe her a special debt of gratitude. We also thank David Siele, director of Human Capital for MDNKOAL, for detailed comments on the sections of the report dealing with education.

Several staff members at the World Bank office in were extremely generous in providing time and tech- nical advice on data analysis. They include Fred Owegi, Frederick Masinde Wamalwa and Catherine Ngumbau. John Mugo, the coordinator of Uwezo Kenya, helped to guide us through the data on learning achievement, and Ibrahim Hussein, the chairman of the Teachers Service Commission at the time of our research, made available technical staff to support our work.

We were extremely fortunate in having the opportunity to present the report to a two-day retreat of policymak- ers, researchers and civil society representatives held in , Kenya. Convened jointly by the Office of the Prime Minister, MDNKOAL, and the Commission on Revenue Allocation (CRA), and facilitated by the United Nations Millennium Campaign, the event produced detailed and constructively critical comments, as well as a lively debate. For feedback and discussion during and after the retreat from the CRA, we thank Micah Cheserem (chair), Fatuma Abdikadir (vice-chair), and Amina Ahmed. We have benefited also from advice on international experience in revenue-sharing from Pinaki Chakraborty of the National Institute of Public Finance and Policy in New Delhi, , and from Dr. Meheret Ayenew of the Forum for Social Studies, . Charles Abugre from the United Nations Millennium Campaign also provided insightful comments.

Research for the report was conducted under the auspices of the Center for Universal Education (CUE) at the Brookings Institution. We were very fortunate to get detailed comments from several colleagues, including CUE Director Rebecca Winthrop. Other CUE staff provided helpful editorial advice, including Anda Adams, Lauren Greubel and Katie Smith. Finally, we thank the William and Flora Hewlett Foundation for supporting the research.*

* The Brookings Institution is a private non-profit organization. Its mission is to conduct high-quality, independent research and, based on that research, to provide innovative, practical recommendations for policymakers and the public. The conclusions and recommendations of any Brookings publication are solely those of its author(s), and do not reflect the views of the Institution, its management, or its other scholars.

Brookings recognizes that the value it provides is in its absolute commitment to quality, independence and im- pact. Activities supported by its donors reflect this commitment and the analysis and recommendations are not determined or influenced by any donation. Contents

Executive Summary...... 1

Introduction ...... 5

The 2010 Constitution: Putting Equity on the Agenda...... 8

Poverty and Health in the 12 Counties...... 15

The 12 Arid and Semi-Arid Land Counties ...... 15

Income Poverty and Inequality ...... 18

Health and Nutrition...... 21

Education: Access and Learning...... 27

The National Picture: School Participation and the Quality of Education ...... 27

The National Learning Achievement Deficit ...... 33

Education Disadvantage in the 12 ASAL Focus Counties...... 39

Some Lessons from International Experience...... 53

Intergovernmental Transfers ...... 53

International Experience...... 54

Social Protection and Safety Nets...... 61

Equitable Sharing for Kenya: The Education Sector and Beyond...... 64

An Initial Framework: The Commission on Revenue Allocation...... 64

Currently Devolved Funds...... 66

Towards Equitable Financing in Education...... 68

The 12 ASAL Counties...... 70

Equitable Financing is Not a Subsitute for Effective and Equitable Policies...... 75

Conclusion...... 77

References ...... 80 LIST OF Figures

Figure 1: Poverty Incidence and Poverty Gap Ranking: 12 ASAL Counties ...... 20

Figure 2: Poverty and Population: County Shares of National Poverty Gap and Population. . . . 21

Figure 3: For Richer, for Poorer: Share of Population in Top and Bottom Quintile of the Wealth Distribution...... 22

Figure 4: The Nutritional Status of Kenya’s Children: Extreme Stunting and Underweight-for-Age...... 24

Figure 5: Progress in Education: Male and Female Enrollment Rate (2000-2009)...... 28

Figure 6: The Age Profile in Kenya’s Classrooms: Age-by-Grade Enrollment (2010)...... 29

Figure 7: Kenya’s Wealth Gaps in School Attendance (2008)...... 30

Figure 8: Charting Grade Progression: Reported Enrollment for Standard 1 Through the KSCE (2003)...... 31

Figure 9: Kenya’s Primary School Learning Outcome Results: National Frequency Distribution for Test Scores (2010)...... 34

Figure 10: Female Literacy and Gender Disparity...... 39

Figure 11: Primary School Ranking: Primary School Net Enrollment Rates and Gender Parity Ratios ...... 41

Figure 12: Secondary School Ranking: Secondary School Gross Enrollment and Gender Parity...... 42

Figure 13: Kenya’s Unequal Distribution of Out-of-School Children ...... 43

Figure 14: School Progression Profiles: Enrollment Levels by Grade for , Turkana, West Pokot and (2009) ...... 44

Figure 15: Unequal Opportunity: Share of County Secondary School-Age Population Sitting Kenya Certificate Secondary Examination (KCSE) (2010) ...... 46

Figure 16: Distribution of Kenya Certificate Primary Examination (KCPE) Test Scores: Turkana, West Pokot, and (2010)...... 48

Figure 17: Unequal Achievement: Learning Levels for Arid Districts and Average for All Districts (2011)...... 50

Figure 18: Actual Constituency Development Fund (CDF) Allocations (2009-2010) vs . Inversion of the Current Formula...... 66

Figure 19: CDF Budget Allocation Including Weighting for Population Density ...... 67 Figure 20: Derived County-Level Share of Primary Education spending as a Proportion of School-Age Population...... 70

Figure 21: Derived Share of FSE Spending as a Proportion of School-Age Population...... 71

Figure 22: Estimated Primary Education Budget Allocations by County: Equal Weighting Attached to School-Age Population and Children in School...... 72

Figure 23: Estimated County-Level Budget Allocations: Current Position vs . A Reinforced Equity Scenario ...... 73

Figure 24: Secondary Education Allocations Under an Equity-Based Financing Formula: Comparison with Current Allocations...... 74

List of Tables

Table 1: Population Size and Share: 12 ASAL Counties...... 17

Table 2: Immunization and Qualified Medical Assistance at Birth: Ranking of 12 Counties and National Average...... 25

Table 3: Maternal Education and Wider Development Indicators ...... 27

Table 4: Reported School Attendance and Enrollment...... 30

Table 5: Kenya Certificate Primary Examination Average Scores: 12 ASAL Counties and National Average (2010) ...... 47

Table 6: KCSE Results: Selected ASAL Counties and National Average (2010)...... 49


Kevin Watkins Woubedle Alemayehu

Executive Summary ‘equitable sharing’ rules governing public spending will provide a litmus test of whether the principles of he constitution adopted by Kenya in 2010 is a re- the new constitution are being translated into public Tmarkable document. It shifts the locus of political policy. authority towards devolved governments, establishes a range of social and economic rights and includes a This paper looks at the implications of the constitu- bill of rights. The constitution also sets out some prin- tion’s public spending provisions for 12 Arid and Semi- ciples on public spending. Future governments will be Arid Land (ASAL) counties.1 Identified by the Ministry required to ensure that revenue and budget alloca- of State for Development of Northern Kenya and Other tions meet constitutional requirements for ‘equitable Arid Lands as areas of specific concern on the basis sharing’. These requirements include affirmative ac- of their deeply entrenched patterns of disadvantage, tion aimed at reducing disparities between regions, these counties have some of the worst social indica- combating marginalization, and raising the quality tors in Kenya. Poverty incidence in several counties and coverage of basic service provision in areas that reaches levels in excess of 80 percent. Poverty in the are lagging behind. ASAL counties is also more intense. While the national poverty gap for Kenya is 16 percent, there are seven These are bold aims. Kenyan society is fractured by ASAL counties in which it is over 30 percent. Food se- deeply entrenched vertical and horizontal inequali- curity problems are endemic. Access to basic services ties, which have been perpetuated and reinforced is limited, with coverage levels far below those in other over time by public spending patterns that system- counties. Drawing on a range of sources, we map depri- atically disadvantaged some groups and areas. The vation in some key dimensions of human development arid and semi-arid regions stand out as areas of acute and rank the counties on a national performance scale marginalization. Human development indicators for covering all 47 of the new counties. these regions fall far below the national average, with populations facing high levels of poverty, food inse- Education is highlighted as an area of special concern. curity and deprivation in access to basic services. It As in other areas, Kenyan society is marked by deep follows that the treatment of these counties under the

financing for a fairer , more prosperous kenya 1 disparities in opportunities for education. These dis- counties and wider inequalities. However, translating parities are closely associated with wider inequalities constitutional principle into public spending strate- linked to poverty, gender and location. Here, too, the 12 gies is not straightforward. The Kenyan government ASAL counties covered in this paper are sites of acute needs to consider how much weight to attach to spe- disadvantage, as illustrated by the following facts: cific disadvantages, the availability of data, and the balance sheet of potential winners and losers. Looking • With 18 percent of Kenya’s primary school-age chil- beyond the considerable technical difficulties in all dren, the ASAL counties account for 46 percent of the out-of-school population. of these areas, any reform of public spending will be shaped by institutional processes that reflect the rela- • Fewer than 10 percent of the children in most of the tive strength—and weaknesses—of different political ASAL counties that we cover negotiate the journey constituencies in shaping decisions. from school entry through the last grade of primary school. The debate over the constitutional provisions on equi- • When Kenya’s 47 counties are ranked by the ratio of table sharing raises a question that goes to the heart of girls-to-boys in primary school, the ASAL counties national governance in Kenya—namely, who gets what account for 11 of the 13 counties with the greatest from the national budget? Answers to that question gender gaps. will be determined by the formulae that govern budget • The ASAL counties account for just 8 percent of allocations. Focusing on horizontal disparities between candidates sitting the Kenya Certificate of Primary counties, we consider a range of options that could Education and just 4 percent of those sitting for the help to translate constitutional principle into budget Kenya Certificate of Secondary Education. practices. Among the central recommendations: • Girls in the 12 ASAL counties are less likely to sit the • The constitution’s ‘equitable sharing’ provision secondary school leaving exam and far less likely should apply to all public spending, and not just than boys to secure a higher grade. Female students the devolved budgets. The distinction is important. are 40 percent less likely to secure a B+ grade or Devolved financing will cover up to 15 percent of higher. national revenue, with a further 0.5 percent of rev- enue allocated through an Equalization Fund. Much The state of education in the 12 ASAL counties has a of the debate in Kenya has tended to focus on these wider significance. If Kenya is to accelerate progress areas. However, the constitution requires that “all towards the 2015 Millennium Development Goal (MDG) aspects of public finance…promote an equitable target of universal primary education and raise aver- society” ( 2011a). Both the let- age learning achievement, government will need to ter and the spirit of its public spending provisions pay far more attention to the most marginalized coun- require that government extends the ‘equitable sharing’ principle to the 85 percent of the budget ties. Failure to tackle education inequalities in general falling outside of the more narrowly-defined de- and the marginalization of children in the ASAL coun- volved funds. ties in particular is acting as a powerful brake on prog- ress towards national goals. • The government of Kenya should adopt a redis- tributive approach to public spending aimed at equalizing opportunity. There are clearly limits to Moves towards greater equity in public spending could the implicit marginal rate of tax that can be applied help to redress the marginalization of the 12 ASAL to finance transfers from richer to poorer counties.

2 Global Economy and Development Program However, current inequalities in access to basic recommend that consideration be given to a de- services, opportunities for education and economic volved budget formula that attaches a weight of 30 infrastructure are a barrier to economic growth and to 50 percent to the poverty gap, as indicated by efficiency, as well as a source of inequality— and the share of each county in the total gap. there is no inherent trade-off between equity on the one side and economic growth on the other. • More weight should be attached to the number of Linked to the right policy framework, more progres- out-of -school children of primary school age and sive public spending has the potential to create a to wider indicators of disadvantage in determin- win-win scenario for equity and economic growth. ing basic education budget allocations. Current Experience from other countries—including more education financing norms allocate resources financially devolved states like India and Brazil— almost entirely to reflect numbers of children in demonstrate intragovernment fiscal transfer has school. This has a perverse unintended effect: coun- helped to mitigate horizontal inequalities without ties with lower levels of school participation are pe- compromising economic growth. nalized. Most of the 12 ASAL counties that we cover lose out. For example, Turkana receives less than • Public spending formulae should reflect a needs- one-half of the public financing for education that based approach to equitable sharing, striking a it would receive if resources were allocated on a per balance between equal per capita transfers and child basis. Current norms for budget transfers at- weighting for disadvantage. Some policymakers tach at best a marginal weight for other indicators in Kenya see equitable financing as synonymous that influence opportunities for education, including with equal per capita transfers. The Commission for poverty, parental literacy, gender, livelihood patterns Revenue Allocation (CRA) has proposed that 60 per- and wider social factors. We advocate a formula- cent of the devolved budget should be allocated on based approach that attaches more weight to (i) the basis of equal per capita transfers. This is an ex- the total number of school-age children in a county tremely narrow and partial interpretation of the term (ii) the poverty gap and (iii) broader indicators of ‘equity’—and one that rests uneasily with the provi- deprivation, including gender disparities. The paper sions of the new constitution. If the aim is to narrow recommends that consideration be given to an ap- inequalities in access to basic services and support proach that weights in the following ranges: 50 per- affirmative action for marginalized counties, as re- cent for children in school, 20 percent for children quested by the constitution, then public spending al- not in school, 20 percent for household poverty, 5 locations have to be positively associated with need percent for gender disparity and 5 percent allocated —that is, the greater the degree of disadvantage, the to a special fund for arid and pastoralist counties. higher the level of support provided. • Secondary education funding formulae should be • The poverty gap, as distinct from the pov - revised to reflect the acute disadvantages facing erty headcount and incidence, should be a the 12 ASAL counties. Kenya’s secondary educa- primary indicator of disadvantage. Reflecting tion budget is increasing both in real terms and as the broader preference for an ‘equal transfer’ a share of the overall education budget. With few model, initial proposal from the CRA in February children—especially girls—in the ASAL counties 2012 recommended an equal cash transfer for progressing beyond secondary school, there is a every person below the national poverty line. danger that budget allocations will increasingly mir- This is a flawed starting point because it ignores ror horizontal inequalities in opportunity. We rec- the depth of their poverty. The ‘poverty gap’ is a ommend a formula-based approach that attaches more sensitive indicator of poverty-related dis - more weight to out-of-school secondary school-age advantage since it captures the distance of the children, with an additional 10 percent of budget average poor person from the poverty line. We allocations weighted to reflect gender disparities.

financing for a fairer , more prosperous kenya 3 • More equitable financing formulae in educa- wide range of social indicators, including the Kenya tion should be linked to more effective policies National Bureau of Statistics and relevant line min- for expanding access and improving learning istries. There are problems however. Surveys on key achievement. The ASAL counties covered in this human development indicators are intermittent. report have some of the lowest levels of participa- In some areas the available data is either dated or tion in education and some of the worst learning not available on a comparable cross county basis. achievement levels in Kenya. More equitable pub- In others, the data is not available. The bottom line lic spending could help to change this picture, but is that policymakers currently lack access to the additional finance has to be linked to policies that type of reliable, real-time data required to inform deliver results. Cash transfers to protect vulnerable approaches to equity. If one of the criteria for ‘eq- households from economic shocks and promote uitable sharing’ is an allocation of resources that demand for education, bursaries for girls, increased reflects need, it is important to develop statistical funding for teacher recruitment and retention, and systems that capture relative deprivation in a timely early childhood programs all have a role to play. The and systematic fashion geared towards annual bud- pastoralist and nomadic livelihood patterns of many geting cycles. people in the ASAL counties also require innovative approaches to education delivery, including invest- The debate over equitable sharing in public finance ment in distance learning and mobile schools. Like is of critical importance for Kenya. The debate, like counties across Kenya, the 12 ASAL counties would the new constitution itself, reflects a growing aware- benefit enormously from a functioning learning as- ness that the country’s extreme disparities in wealth sessment system, improved teacher training and and opportunity represent not just a source of social strengthened in-service support. injustice, but a barrier to economic efficiency, shared • Looking to the future, policymakers in Kenya growth and political stability. The Kenyan govern- should consider the development of a public fi- ment’s broad strategy for the future of the country, set nancing model geared towards the provision of out in Kenya Vision 2030, aims at the development of a a ‘social minimum’ of basic services. Through its national commitments to the Millennium socially inclusive society and a more competitive econ- Development Goals and the social and economic omy. Neither of these aims is compatible with the ex- rights enshrined in the constitution, Kenya has com- treme inequalities that now characterize the country. mitted to provide a basic standard of provision for all citizens. As the devolved system develops over For many Kenyans, devolution and the provisions of time, county-level and national authorities should the new constitution hold out the promise of a more estimate the financing gap facing each county with equitable pattern of development. Public spending respect to the provision of key basic services. That gap should figure as a ‘needs assessment’ compo- has a critical role to play in realizing that promise. nent in national and devolved financing formulae. Spreading opportunity and investments more equally across the country could unlock the door to acceler- • The government of Kenya and aid donors should ated human development, reduced inequality and ac- invest in building the national statistical capac- celerated growth. Making the transition to equitable ity required to underpin a devolved financing system and to inform approaches to equitable sharing will not be easy—but a business-as-usual al- sharing. Kenya has a relatively strong and profes- ternative is likely to prove politically and economically sionalized set of institutions generating data on a unsustainable.

4 Global Economy and Development Program Introduction also on an upward trend. On most measures of hu- man development, Kenya registers average outcomes n August, 2010 the government of Kenya adopted considerably above those for sub-Saharan as Ia new constitution. This followed a referendum in a region. Yet the national average masks extreme which an overwhelming majority of Kenyans voted disparities—and the benefits of increased prosperity for change. The decisive impetus for reform came have been unequally shared. Some regions and social from the widespread violence and political crisis that groups face levels of deprivation that rank alongside followed the 2007 election. While claims of electoral the worst in Africa. Moreover, the deep fault lines run- fraud provided the immediate catalyst for violence, ning through society are widely perceived as a source the deeper causes were to be found in the interac - of injustice and potential political instability. tion of a highly centralized ‘winner-take-all’ political system with deep social disparities based in part on High levels of inequality in Kenya raise wider con- group identity (Hanson 2008). cerns. There has been a tendency in domestic debates to see ‘equitable sharing’ as a guiding principle for Provisions for equity figure prominently in the new social justice, rather than as a condition for acceler- constitution. Backed by a bill of rights that opens the ated growth and enhanced economic efficiency. Yet door to legal enforcement, citizenship rights have international evidence strongly suggests that extreme been strengthened in many areas,including access inequality—especially in opportunities for educa- to basic services. ‘Equitable sharing’ has been in- tion—is profoundly damaging for economic growth. troduced as a guiding principle for public spending. It follows that redistributive public spending has the National and devolved governments are now consti- potential to support growth. tutionally required to redress social disparities, target disadvantaged areas and provide affirmative action The current paper focuses on a group of 12 counties for marginalized groups. located in Kenya’s Arid and Semi-Arid Lands (ASALs). They are among the most disadvantaged in the coun- Translating these provisions into tangible outcomes try. Most are characterized by high levels of income will not be straightforward. Equity is a principle that poverty, chronic food insecurity and acute deprivation would be readily endorsed by most policymakers in across a wide range of social indicators. Kenya and Kenya’s citizens have provided their own endorsement through the referendum. However, there Nowhere is the deprivation starker than in education. is an ongoing debate over what the commitment to The ASAL counties account for a disproportionately equity means in practice, as well as over the pace and large share of Kenya’s out-of-school children, pointing direction of reform. Much of that debate has centered to problems in access and school retention. Gender on the constitutional injunction requiring ‘equitable disparities in education are among the widest in the sharing’ in public spending. country. Learning outcomes for the small number of children who get through primary school are for the There are compelling grounds for a strengthened fo- most part abysmal, even by the generally low national cus on equity in Kenya. In recent years, the country average standards. has maintained a respectable, if less than spectacu- lar, record on economic growth. Social indicators are

financing for a fairer , more prosperous kenya 5 Unequal public spending patterns have played no Part 2 provides an analysis of some key indicators on small part in creating the disparities that separate the poverty, health and nutrition. Drawing on household ASAL counties from the rest of Kenya—and ‘equitable expenditure data, the report locates the 12 ASAL sharing’ could play a role in closing the gap. But what counties in the national league table for the incidence would a more equitable approach to public spending and depth of poverty. Data on health outcomes and look like in practice? access to basic services provide another indicator of the state of human development. While there are This paper addresses that question. It looks in some some marked variations across counties and indica- detail at education for two reasons. First, good qual- tors, most of the 12 counties register levels of depriva- ity education is itself a powerful motor of enhanced tion in poverty and basic health far in excess of those equity. It has the potential to equip children and youth found in other areas. with the skills and competencies that they need to break out of cycles of poverty and to participate more Part 3 shifts the focus to education. Over the past fully in national prosperity. If Kenya is to embark on decade, Kenya has made considerable progress in im- a more equitable pattern of development, there are proving access to basic education. Enrollment rates in strong grounds for prioritizing the creation of more primary education have increased sharply since the equal opportunities in education. Second, the educa- elimination of school fees in 2003. Transition rates tion sector illustrates many of the wider challenges to secondary school are also rising. The record on and debates that Kenya’s policymakers will have to learning achievement is less impressive. While Kenya address as they seek to translate constitutional provi- lacks a comprehensive national learning assessment, sions into public spending strategies. In particular, it survey evidence points to systemic problems in edu- highlights the importance of weighting for indicators cation quality. In both access and learning, children in that reflect need in designing formulae for budget al- the ASAL counties—especially female children—are locations. at a considerable disadvantage. After setting out the national picture, the paper explores the distinctive The paper is organized as follows. Part 1 provides an problems facing these counties. overview of the approach to equity enshrined in the constitution. While the spirit of the constitution is un- In Part 4 we look beyond Kenya to wider interna- equivocal, the letter is open to a vast array of interpre- tional experience. Many countries have grappled tations. We briefly explore the implications of a range with the challenge of reducing disparities between of approaches. Our broad conclusion is that, while less-favored and more-favored regions. There are no Kenya clearly needs to avoid public spending reforms blueprints on offer. However, there are some useful that jeopardize service delivery in wealthier counties, lessons and guidelines that may be of some relevance redistributive measures are justified on the grounds to the policy debate in Kenya. The experience of South of efficiency and equity. Although this paper focuses Africa may be particularly instructive given the weight principally on basic services, we caution against ap- attached to equity in the post-apartheid constitution. proaches that treat equity as a matter of social sector financing to the exclusion of growth-oriented produc- Part 5 of the paper explores a range of approaches tive investment. to financial allocations. Converting constitutional

6 Global Economy and Development Program principle into operational practice will require the de- or inputs. These questions go beyond devolved financ- velopment of formulae-based approaches. From an eq- ing. The Kenyan constitution is unequivocal in stipu- uitable financing perspective there is no perfect model. lating that the ‘equitable sharing’ provision applies to Any formula that is adopted will involve trade-offs be- all public spending. We therefore undertake a series tween different goals. Policymakers have to determine of formula-based exercises illustrating the allocation what weight to attach to different dimensions of equity patterns that would emerge under different formulae, (for example, gender, income, education and health), with specific reference to the 12 ASAL focus counties the time frame for achieving stated policy goals, and to education. and whether to frame targets in terms of outcomes

financing for a fairer , more prosperous kenya 7 The 2010 Constitution: civil and political rights. While there are potentially Putting Equity on the formidable social, economic and legal barriers fac- Agenda ing any citizen seeking redress through litigation, the newly established Kenya National Human Rights he 2010 constitution marks the most momen - and Equality Commission is charged with promoting Ttous governance reform in Kenya’s post-indepen- and protecting the rights set out in the constitution, dence history (Akech 2010; Kramon and Posner 2011). including social and economic rights (Kenya National Devolution is at the heart of the reform process. While Commission on Human Rights 2011). many of the details on implementation remain unclear, the new constitution signals a shift in the locus of power away from a highly centralized system and towards Public Spending and ‘Equitable Sharing’ decentralized government at the county level. It also Public finance figures prominently in the social pro- establishes social justice as a guiding principle for policy visions of the new constitution. Chapter 12 sets out design. Drawing on the experience of South Africa, the guidelines establishing equity as an organizing princi- constitution includes far reaching provisions aimed at ple for the allocation of public spending. These guide- making Kenya a fairer and more equal country. lines include an injunction to ensure that ‘expenditure shall promote the equitable development of the coun- Devolution will transform the structure of government. try, including by making special provision for margin- An explicit objective is to bring decision making closer alized groups and areas’ (Government of Kenya 2011a) to the affected population and to make government more accountable. Under the old constitutional regime, That injunction extends beyond the devolved budgets. political power was highly concentrated in central gov- Article 202 stipulates that ‘revenue raised nationally ernment with administration structured around eight shall be shared equitably among national and county provinces and 158 districts. The new devolved system governments.’ Criteria to be used in allocating fund- will operate through 47 counties. The new counties, ing include ‘economic disparities within and among which will become operational during 2013 after the counties and the need to remedy them,’ as well as next election, will have responsibility for delivering a ‘the need for affirmative action in respect of disad- wide array of ‘proximate, easily accessible services’, vantaged areas and groups.’ This is not the first time promoting ‘the interests and rights of marginalized that equity has been made a budget priority. Indeed, communities’ and overseeing the ‘equitable sharing’ of in a review of past public spending practices one com- resources in their areas of assigned responsibility. mentator has observed that ‘at the planning stage, inequality is a priority but there is no link between Equity is the political cornerstone of the new constitu- plans and budgets’ (Kiringai 2006). Where the new tion. The longest chapter is a Bill of Rights, the provi- constitution is distinctive is in the scope and, poten- sions of which can be limited only under exceptional tially, enforceability of its provisions. circumstances, creating a constitutional framework for legal challenges to government policies (Domingo While the constitutional commitment to equity is and Wild 2012). The constitution creates quasi-legal unequivocal wider principles also apply. Article 203 entitlements to basic services as a right of citizen- establishes 11 separate criteria for determining equita- ship, along with a broad array of social and economic,

8 Global Economy and Development Program ble shares, including national interest, fiscal capacity, 2011a). Another 0.5 percent of revenue will be chan- efficiency, development needs, the ‘economic opti- neled through the Equalization Fund, which the mization of each county,’ and developmental needs. constitution requires be used ‘to provide basic ser- While uncontroversial in themselves, these provisions vices … to marginalized areas to the extent necessary underline the potential for divergent interpretation. to bring the quality of those services … to the level It is not difficult to envisage a scenario in which ad- generally enjoyed by the rest of the nation, so far as vocates for ‘economic optimization’ seek to assert possible.’ This language is important because it estab- precedence over those calling for more weight to be lishes an explicit requirement that spending is geared attached to ‘economic disparities … and the need to not just towards expanded provision of basic services, remedy them.’ Similarly, claims from county govern- but also towards a reduction of inequalities in the level ments prioritizing affirmative action may be met by and quality of provision. The fund can also be used by counterclaims from those highlighting fiscal capacity. the central government to provide conditional grants Constitutional documents do not, by their nature, pre- or other direct financing to counties with marginal- scribe detailed resolutions for potential conflicts over ized populations. interpretation and the new Kenyan constitution is not an exception to this rule. Transition arrangements for the two major devolved funds now in operation have yet to be finalized. Currently, the main source of decentralized financ- Devolved Financing and Beyond ing in Kenya is the Local Authorities Transfer Fund Since the adoption of the constitution, much of the (LATF), a block grant provided by central government debate on equitable financing has focused on de- and used at the local authority’s discretion. Equity volved budgets. This may be misplaced on two counts. considerations of the type set out in the new consti- First, the bulk of public spending—over 80 percent of tution play a limited role in determining allocations the total—will continue to come from central govern- under the LATF because of the weighting attached ment budgets. It follows that overall equity in public to population and lump sum transfers. 2 The other spending will be shaped by policies influencing the major devolved financing vehicle is the Constituency nondevolved budgets. Second, the new constitution Development Fund (CDF). Established in 2004, the explicitly requires that the principles of ‘equitable legislation governing the CDF requires that it receive development’ and ‘special provision for marginal- 2.5 percent of government revenue for allocation to ized groups and areas’ applies to all public spending development programs at a constituency level. While (Article 201) (Government of Kenya 2011a). This is par- equity weighs more heavily than under the LATF, the ticularly important in light of the fact that some key bulk of CDF financing is allocated on an ‘equal shares’ basic services—including education—will not initially formula with just 25 percent of the transfer linked to be devolved to county-level governments. poverty (Government of Kenya 2010; Romero 2009).

The constitution sets some clear budget parameters. Under the ‘equitable share’ provision counties are Why Equity Matters guaranteed to receive not less than 15 percent of Equity has emerged as an increasingly prominent national revenue (Article 203) (Government of Kenya theme in national policy debates in Kenya. It figures

financing for a fairer , more prosperous kenya 9 with some prominence in the government’s Kenya mortality rates among children from the poorest 20 Vision 2030 strategy, which seeks to identify a path- percent of homes are twice as high as they are among way to middle-income status and the development of the wealthiest households (Kenya National Bureau a fairer, more inclusive society (Government of Kenya of Statistics 2010). The most recent service delivery 2007). The emphasis on fairness and inclusive growth survey found that only 56 percent of clinics in North in Kenya Vision 2030 reflects wider currents of thinking offered antenatal care, compared to in international development. In recent years reports 94 percent in . from the World Bank and the Africa Progress Panel These disparities in wealth and opportunity are in di- have drawn attention to the damaging consequences rect conflict with many of the goals set out in national of extreme inequality not just for social justice, but also policy documents, including Kenya Vision 2030: for political stability, economic growth and poverty re- duction (World Bank 2006; Africa Progress Panel 2012). • Poverty reduction. High levels of inequality weaken the rate at which economic growth is con- Efforts to promote equity are in some senses swim- verted into poverty reduction—the poverty elastic- ming against the tide of post-independence history. ity of growth. Other things being equal, increasing the share of increments to growth captured by Enduring inequalities in Kenya reflect the legacy of an people living in poverty will accelerate the pace of uneven pattern of economic growth, unequal provi- poverty reduction (Ravallion 2005; Ravallion 2009; sion of basic services, and a political system that has Ferreira and Ravallion 2009). The effect is cumula- perpetuated group-based disparities in political power tive because poverty reduction is itself a potential (Muhula 2009; World Bank 2009). Income distribution spur to increased investment, productivity and eco- is highly unequal, with the Gini coefficient estimated nomic growth. Kenya’s variable record in converting growth into poverty reduction illustrates the impor- at 0.44—well above the level in neighboring countries tance of patterns of distribution (Kabubo-Mariara, such as Ethiopia, and . Economic Mwabu and Ndeng’e 2012). growth has been skewed towards urban centers, a narrow corridor between the port of and • Economic growth. It has sometimes been argued , and a small number of commercial farming that pro-poor redistribution is counterproductive for poverty reduction because it has the potential areas. According to the Word Bank, 80 percent of eco- to damage economic growth. In practice, outcomes nomic activity is generated by just half of Kenya’s new will depend on the design, pace and sequencing of counties (World Bank 2011b). reforms. However, there is now a large and grow- ing body of evidence to suggest that the implied Wealth disparities intersect with wider inequalities. trade-off between growth and equity is more imag- While the average Kenyan aged 17-22 years has spent ined than real (Bourguignon, Ferreira, and Walton 2007; World Bank 2006). Recent analysis from the just over seven years in school that figure rises to International Monetary Fund and others indicates over 10 years for the wealthiest 20 percent. Similarly, that high levels of inequality are bad for long-term around 12 percent of 17-22 year olds have accumulated economic growth and poverty reduction (Berg and less than four years in school, rising to 27 percent Ostry 2011). In the case of Kenya, income inequal- for girls from poor rural households and 92 percent ity and inequalities in opportunities for health, for girls from the ethnic Somali community (UNESCO education and nutrition compromise the economic 2010). Health disparities are equally marked. Child growth goals set out in Kenya Vision 2030.

10 Global Economy and Development Program • Social cohesion. High levels of inequality in income Fiscal pressure provides another rationale for and opportunity can weaken political institutions strengthening the commitment to equity. While gov- and exacerbate tensions between groups (Alesina ernment spending has increased strongly in real and Rodrik 1994; World Bank 2006; Fukuyama terms since 2003, the slowdown in economic growth 2012). This is one of the reasons that the new con- stitution prioritizes enhanced equity. As the Kenya and the stimulus package adopted in the 2009/2010 Vision 2030 document recognizes, perceived injus- budget has led to a deteriorating budget position. tices and disparities in access to basic services have With the fiscal deficit nearing 6 percent of GDP and been “a major cause of social tensions in the coun- interest payments taking a rising share of the recur- try as was evident during the 2007 post-election rent budget, public spending is likely to increase only crisis” (Government of Kenya 2007). slightly over the next few years. Reducing poverty and • Lost human potential. Extreme inequalities in enhancing equity while adhering to the priorities set opportunity come with high costs for individuals out in the medium-term expenditure framework will and society. For example, education is a strong pre- require greater attention to equity in public spending. dictor of individual earnings and is also strongly correlated with health status, which in turn influ- The 2010 constitution itself provides little guidance ences earnings. Cross country evidence from rich on what this might mean in practice. At one end of the and poor countries suggests that gains in educa- tion quality can raise the long-run average annual spectrum, equity might be interpreted as a require- growth rate by 2 percent per capita, with attendant ment that all Kenyans receive an equivalent level of benefits for poverty reduction (Hanushek 2009; financing. Other approaches might place more em- Brown 2011). Thus greater equality of opportunity phasis on the equalization of opportunity. Recognizing for education can help to promote not just human that some populations may require more funding to development but more efficient and more dynamic secure equivalent opportunities as a result of, say, economies (World Bank 2006). poverty or illness, more equitable public spending Greater equity would enhance Kenya’s prospects could be seen as a matter of matching resources with for accelerated progress towards the Millennium needs, to achieve equivalent capabilities—an aspect Development Goals. While trend data is lacking, World of equity to which Amartya Sen drew attention (Sen Bank estimates suggest that 45 percent of Kenyans 1992). Applied in the context of an intragovernmental were living on less than $1.25 a day in 2005, above revenue transfer system, a greater emphasis on the the estimated levels for 1990 (the MDG base year). equalization of opportunity would require financing According to the 2010 Public Expenditure Review, formulae geared towards the correction of horizontal progress in reducing poverty may have slowed since and vertical disparities in opportunity linked to wider 2008. Scenarios for under-five mortality, maternal disadvantages. To provide a practical example, more mortality and access to water have also deteriorated equitable financing for basic education would require with the 2010 Public Expenditure Review estimating not just spending on education infrastructure and that the amount of finance required to achieve the teachers in underperforming areas, but additional per MDGs had increased from $49 per capita (in 2005) to capita transfers to counteract the effects of disadvan- $68 per capita (Government of Kenya 2010). tages transmitted through poverty, malnutrition and parental illiteracy.

financing for a fairer , more prosperous kenya 11 In practice, all public spending systems have to strike inequalities, with strategies for inclusive economic a balance between ‘population-based’ and ‘needs- growth. China is a case in point. During the 1990s based’ transfers. Depending on the perspective concern over regional inequalities prompted the adopted, there are a range of approaches to public Chinese government to introduce a program— the spending and service provision that could claim some ‘8-7 Program’—targeting the poorest counties in degree of consistency with constitutional principles the country for increased investment in productive in favor of ‘equitable sharing’. Most national revenue- infrastructure and enterprises. The result was a sig- sharing models would include some or all of the fol- nificant increase in growth and employment (Higgins, lowing elements (Bahl and Linn 1994; Bahl 2008): Bird and Harris 2010). Since 2000, there has been a renewed emphasis on investment in poorer regions. • Equal per capita transfers based on population, In Vietnam, another high growth economy, ‘Program irrespective of differences in wealth, location or relative need. 135’ targeted 2,374 poor communes in ethnic and mi- nority areas not just for social service provision, but • Equal share transfers under which each county re- also with productive investment in roads, irrigation ceives a fixed share of a specified budget. and the development of markets. The program is cred- • Deprivation-weighted transfers under which ited with supporting economic growth and poverty disadvantaged groups identified by, say, poverty, reduction (Thuat and Quan 2008). Social protection health indicators and other sources of disadvantage is another example of a redistributive public spend- receive a budget increment. ing policy with the potential to raise the income and • Outcome-based transfers under which budgets productivity of the poor. In Brazil, the Bolsa Familia are allocated to reflect commitment to a specific cash transfer scheme has transferred around 0.4 result, such as reduced disparities in access to and percent of national GDP to the poorest households utilization of education, health care and other basic in the country. The program is overtly redistributive. services, or a specific goal such as universal pri- There is no evidence that it has weakened economic mary education and improved child survival. growth. What it has done is to contribute to a decade • Cost-related transfers that reflect the financing of rapid poverty reduction and falling inequality, with required to deliver basic services in areas charac- the average income of the poorest households rising terized by different population densities, accessibil- at three times the level of the wealthiest households ity and other factors affecting cost. (Ravallion 2009). The Brazilian case is one element in a wider regional story. Over the past decade, redis- There are no blueprints for guiding the design of ap- tributive social protection programs in Latin America proaches to equitable public spending. The World have contributed to a region-wide decline in inequality Bank has drawn attention to the potential for damag- and stronger economic growth (Cornia 2012; ECLAC ing trade-offs between economic growth on the one 2011). Here, too, the evidence is that redistributive side and redistributive public finance on the other equity and growth-oriented policies can be mutually (Box 1). While the report makes a number of impor- reinforcing. tant observations, the central policy prescription to emerge rests uneasily with evidence from countries None of this is to discount potential tensions be- that have sought to combine more equitable public tween efficiency and equity to considerable scrutiny. spending aimed at narrowing horizontal and vertical

12 Global Economy and Development Program Box 1: ‘People Not places’:The World Bank Perspective Second, the World Bank’s perspective takes it as axiomatic that Kenya’s poorer counties have grown more slowly be- Building on an analytical framework developed in the 2009 cause they lack the advantages of high growth areas, and World Development Report, the World Bank has recom - that migration offers the best near-term prospect of more mended that devolution in Kenya should be guided by the inclusive growth. This may be confusing cause and effect. It principle that equalization should target ‘people not places.’ could equally be argued that high growth commercial farm- What does this mean in practice? ing areas have emerged as growth poles in part because of the infrastructure support that they have received, while arid The World Bank’s starting point is that equity should focus and semi-arid areas have grown less rapidly because of weak not on the progressive equalization of household or regional infrastructure. incomes, but on investments aimed at expanding access to education, health and other basic services. The reason: in- There are certainly grounds for concern that successive gov- creasing income disparity during initial growth surges is seen ernments in Kenya and the donor community have under- as an inevitable consequence of the advantages enjoyed by estimated the growth potential of arid and semi-arid lands. economic growth poles. Using the public finance system to Comparisons with Ethiopia, where arid and semi-arid areas narrow wealth gaps has the potential to weaken the very have emerged as a growth pole, are instructive. These areas incentives that drive growth, while at the same time disrupt- are at the center of a livestock and meat sector that is now ing service provision in high growth areas. The preferred op- the second largest exporter after coffee, accounting for 12-15 tion in this perspective is to use the revenues generated by percent of foreign exchange earnings. Linkages with higher growth to progressively strengthen basic service provision. value-added, labor-intensive manufacturing such as foot- As the World Bank puts it: wear and leather are strengthening over time. By contrast, Kenya has been slow to exploit the potential of livestock “Large-scale distribution across counties may not be pos- sectors in arid and semi-arid regions, in part because of a sible or desirable immediately, given budget constraints and restricted assessment of growth prospects. Recent research efficiency considerations...these ‘wealthy’ counties are also put the contribution of livestock to the national economy at among Kenya’s most dynamic regions, which are driving eco- $4.2 billion in 2009, or 13 percent of the total. That is more nomic growth and generating the bulk of national income out than double previous assessments and not far off the $5.25 of which redistribution will eventually be financed….Any dras- billion estimated value of crops and horticulture. There are tic move to redistribute resources away from affluent towards other areas of potential growth in the arid and semi-arid destitute counties could result at best in severe fiscal stress, counties, including renewable energy, minerals, and linkages and at worse in the collapse of essential service delivery.” to neighboring economies.

Stated as a matter of abstract principle, this view is super- The third element of the World Bank’s approach meriting ficially attractive. Expanding service provision in one area critical scrutiny is the starting point. While it is right to cau- through financing arrangements that lead to the collapse tion against an excessively redistributive approach at high of services in another is clearly a suboptimal approach. In implied rates of marginal taxation, the real debate is over the economic terms, an excessive marginal rate of taxation on balance to be struck between the pursuit of equity and real- income and wealth creation is potentially damaging. ism in public finance. Some degree of redistribution is both affordable and desirable. Many would argue that redistribu- Yet it is not immediately clear where equity fits in to the tive public finance is also a political imperative and a consti- logic of the World Bank’s perspective. In the case of Kenya, tutional obligation. Ultimately, policymakers in Kenya have the hope for disadvantaged counties appears to be that re- to design equitable sharing financing policies that reflect distributive transfers will ‘eventually’ take place once some a concern to simultaneously narrow gaps in opportunity, specified wealth threshold has been passed. The underlying mitigate horizontal and vertical inequalities, and support message is that little can be done through more equitable economic growth. sharing of revenues to mitigate Kenya’s deep disparities in access to basic services, and that spending patterns under- Sources: World Bank 2011a; World Bank 2011b; Higgins, Bird pinning those disparities will remain intact for the foresee- and Harris 2010. able future.

financing for a fairer , more prosperous kenya 13 Transfers to poor regions financed by excessive mar- prospects for growth. Ultimately, policymakers need ginal tax rates on wealthier regions and higher income to consider the scale of transfers, the time horizon for groups could have the perverse effect of improving achieving equity goals and the full range of intended equity access in the short run, while undermining the and unintended outcomes. There are self-evidently potential for economic growth, employment creation limits to the extent of feasible redistribution, but rul- and financing for basic services over the medium ing out the scope for redistributive income transfers term. Yet the sides of perverse incentives operating in advance is not necessarily a good guide to policy in the other direction also have to be recognized. formulation. Extreme horizontal disparities can also weaken

14 Global Economy and Development Program Poverty and Health in the 12 population density, the strong influence of clan-based Counties governance systems, and the distinctive livelihood challenges facing pastoralists. n this section we chart some of the key human de- Ivelopment deficits affecting the 12 ASAL counties. While the 12 focus counties included in this report face After identifying the distinctive features of the coun- many development problems in common there are ties, we focus on income poverty, wealth distribution also important differences. Eight of them—Garissa, and a range of health indicators. Disadvantages in Mandera, , Samburu, Tana River, Turkana, each of these areas are important in themselves while and Wajir— are arid and spread over very at the same time symptomatic of wider inequalities large areas. Another two counties— and West in opportunity. Where possible, we locate the 12 ASAL Pokot—are less arid and cover smaller geographic ar- counties on a national scale of disadvantage relative eas. Both and are categorized as semi- to other counties. The horizontal disparities captured arid. Human development indicators also vary widely. in these rankings are one part of a wider picture of As the data presented in this section demonstrates, inequality (Sundet and Moen 2010; World Bank 2009). Kajiado and Narok are outliers. Bordering commercial They intersect with the fault lines running through agricultural areas and in the case of Kajiado, Nairobi, Kenyan society linked to wealth, gender, ethnicity, they figure near the top of the national for some— rural-urban differences and other determinants. Yet although not all—of the indicators that we examine. the ASALs are centers of highly concentrated disad- At the other end of the spectrum, counties such as vantage. With a small number of exceptions, they have Turkana and Wajir are consistently at or around the the highest incidence of poverty and they account for bottom of the national league table. a disproportionate share of the national poverty gap. Kenyans living in the ASAL counties also face acute The limitations of a county-based analytical lens have disadvantages in access to health and education. to be recognized. To the extent that the new coun- While we trace different aspects of deprivation sepa- ties represent the locus of political devolution, cross rately, it is important to recognize that the disadvan- county disparities in human development will have tages they entail operate in a cumulative and mutually to figure prominently in any needs-based financing reinforcing fashion to diminish life-chances. formula aimed at equitable sharing. However, just as national average data can mask disparities across The 12 ASAL Counties counties, so county-level data can obscure inequali- ties within the 12 counties. For example, while Kajiado The counties covered in this report represent a subset and Narok have low levels of income poverty, there of the ASAL counties. Of the 47 new counties created are large variations around the average. The pov - through devolution, 23 are categorized as ASAL ar- erty incidence figure for is 34 percent, eas. The Ministry of Northern Kenya has responsibili- with a reported incidence for the two old districts of ties spanning all 23 counties. However, it has identified Transmara and Narok (which were combined to form the 12 counties that we cover as areas meriting special the new county) were 50 percent and 26 percent re- concern in the context of political devolution. Those spectively. Similarly, in a county like Garissa there are concerns relate to human capital weaknesses, the very large disparities between urban and rural areas size of geographic areas covered and associated low

financing for a fairer , more prosperous kenya 15 Map 1: Kenya’s New County Map with Selected ASAL Counties

Turkana Mandera


Wajir West Pokot Samburu

Trans Nzoia Isiolo Elgeyo Marakwet Baringo Uasin Gishu Busia Laikipia Meru Nandi Kisumu Tharaka Nithi Nyandarua NyeriKirinyaga Garissa Embu Kisii Muranga Migori Narok Nairobi Tana River

Makueni Lamu Kajiado

Kilifi Talta Taveta

Kilometers Mombasa 0 50 100 200 300 400

Source: created by The World Bank: Nairobi 2012.

16 Global Economy and Development Program across all indicators. It follows that any approach to Table 1: Population Size and Share: 12 more equitable sharing in public finance will have to ASAL Counties look beyond county-level indicators to subcounty data Population Population Share and inequalities within the counties. County Size (percent) Garissa 623,060 1.6 As is evident from Map 1, the 12 selected ASAL coun- Isiolo 143,294 0.4 ties dominate the physical . Just five of these counties account for almost half of total Kajiado 687,312 1.8 land area: in descending order, Marsarbit, Turkana, Lamu 101,539 0.3 Wajir, Garissa and Tana River. Collectively, these 12 Mandera 1,025,756 2.7 counties have a population of around 6.2 million Marsabit 291,166 0.8 people, or 16 percent of the national total (Table 1). Narok 850,920 2.2 Low population density is one of the characteristics of Samburu 223,947 0.6 almost all of the 12 counties. Average population den- Tana River 240,075 0.6 sity in Kenya is 66 people per km2, rising to over 4,000 Turkana 855,399 2.2 people per km2 for Nairobi. The comparable figures for the 12 ASAL counties range from 4-6 people per km2 Wajir 661,941 1.7 for Marsarbit and Tana River, to 12-13 people per km2 West Pokot 512,690 1.3 for Wajir and Turkana, and over 30 people per km 2 Total 6,217,099 16.1 for Kajiado and Narok. The 12 focus counties are also Source: Census 2009. home to a large and growing share of Kenya’s young people, with half of the population in the 12 counties and comparability problems. For instance, the most aged 15 years or less. recent data available on poverty and inequality is from the 2005 Kenya Household Budget Survey; and at the time this report was being prepared (late 2011) Whatever definition of equity is adopted, the devel- the 2010 county-level school enrollment data were opment of equitable financing formulae depends still not accessible. If the aim is to ensure that public critically on the availability of credible, recent and spending formulae reflect need, then more has to be relevant data. Our cross county ranking revealed a done to generate real-time data aligned with the bud- number of problems in this area. At one level, Kenya get cycle. Once the key indicators for assessing equity is a ‘data rich’ country. Many government line min - and allocating resources are determined, it is critically istries have statistics units. The Kenya National important that government develops the capacity to Bureau of Statistics publishes a wide range of data. collect and disseminate data on a timely basis. This Household surveys are widely used by government, capacity is needed at both national and county levels. nongovernment organizations, the World Bank and the U.N. to generate data on different parts of the Comprehensive county-level data was not available at country. Additionally, moves towards greater trans- the time that research for this report was undertaken. parency through the Kenya Open Data Initiative have The World Bank has re-estimated data in the 2005 increased both the quantity and the quality of publicly Kenya Integrated Household Budget Survey (KIHBS), available data. Yet there are significant gaps, time lags

financing for a fairer , more prosperous kenya 17 which we draw on for the statistics on income poverty —16.6 million people in 2005— had levels of con- and inequality. The 2005 survey also makes it possible sumption insufficient to meet basic food needs, with to measure at a county level the depth of poverty, as marked differences between urban and rural areas indicated by the income-gap ratio.3 Health and edu- (a reported incidence of 33 percent and 49 percent cation data are drawn from administrative reporting respectively). These figures represented a modest systems, reconfigured to follow the administrative decline over the levels reported in 1997.4 For point of contours of county boundaries. Drawing on these reference, the estimated poverty incidence in 2005 sources the following section provides a snapshot of for the international $1.25 purchasing power parity where the 12 focus counties stand in a national rank- threshold was 43 percent. The KIHBS survey also ing for selected indicators covering poverty, inequality provides an estimate of the poverty gap. Whereas and public health the incidence of poverty measures the share of the population below the poverty line, the poverty gap provides information regarding how far households Income Poverty and Inequality fall from the poverty line.5 It captures the mean aggre- Household expenditure is an important indicator of gate income (or consumption) shortfall relative to the welfare. While monetary wealth is a means to broader poverty line across the whole population. The national ends rather than a direct measure of human capabil- poverty gap for Kenya at the time of the survey was ity, income poverty in Kenya is closely associated with 16.3 percent. Another threshold used in the KIHBS is wide-ranging disadvantages in health, nutrition and ‘hardcore poverty,’ a cut-off line at which individuals education. For all of these reasons, the distribution of would be unable to meet basic food needs even if they household consumption poverty is an important di- were to forego all nonfood consumption. Around one mension of horizontal and vertical inequality in Kenya. in five individuals in that year had consumption levels falling below this line. While hardcore poverty fell sig- As in many other countries in sub-Saharan Africa, the nificantly in rural areas over the decade prior to 2007 poverty data in Kenya suffer from the irregularity of it slightly rose in urban areas. measurement and inconsistencies between house- hold surveys and national income accounts. The most recent national data for Kenya comes from the 2005- Poverty in the ASAL Counties 2006 Kenya Integrated Household Budget Survey, Any approach to equitable sharing in public finance which provides a snapshot of poverty some eight has to consider whether and how to weight for pov- years ago. Inevitably, the national poverty profile will erty. This is not a straightforward exercise. Poverty have changed over the intervening years. The result- can be measured by reference to incidence, headcount ing data gap highlights the need for the development numbers or the poverty gap, all of which provide dif- of more timely data collection, perhaps using smaller- ferent perspectives. In any system of intergovernmen- scale but more frequent surveys to update the picture tal transfer formulae that attach more weight to the provided by large-scale survey exercises. incidence of poverty than the headcount will implicitly favor those regions with high poverty rates over those The KIHBS estimated poverty by reference to a range with lower rates but larger numbers of poor people. of thresholds. It reported that 46 percent of Kenyans One of the limitations of both the incidence and the

18 Global Economy and Development Program headcount measures is that neither captures the differences across the 12 counties. Poverty incidence depth of poverty. The poverty gap is in many ways the in Mandera, Marsabit, Turkana and Wajir is over twice most useful sensitive indicator of poverty-related dis- the national average, reaching 94 percent in Turkana. advantage, since it combines the headcount number On the national ranking, these four counties, along with the distance from the poverty line. with Samburu, Tana River and West Pokot, account for seven of the 10 counties with the highest incidence Both the incidence and the depth of poverty are far of poverty in the country (Figure 1). Poverty is also higher in most of the 12 ASAL counties than in the deeper in the ASAL counties (Figure 1). The reported rest of Kenya. Weighted poverty incidence for the poverty gap for Mandera was 44 percent, rising to counties averaged 61 percent, with some 2.5 million 69 percent for Turkana. This implies that the aver- people affected. Averages inevitably obscure the age income of a poor person in Turkana is less than

Figure 1: Poverty Incidence and Poverty Gap Ranking: 12 ASAL Counties

100 Poverty Incidence





0 Kisii Kitui Meru Siaya Busia Embu Nandi Kwale Vihiga Migori Bomet Nairobi Nakuru Kisumu Kericho Kiambu Baringo Laikipia Nyamira Makueni Lamu (6) Kirinyaga Murang'a Bungoma Narok (9) Mombasa Wajir (45) Wajir Homa Bay Machakos Kajiado(1) Isiolo (36) Kakamega Nyandarua Trans Nzoia Trans Taita Taveta Taita Uasin Gishu Garissa (29) Turkana (47) Turkana Tharaka Nithi Marsabit (44) Mandera (46) Samburu (43) Tana River (42) Tana West Pokot (40) Pokot West National Average Elgeyo/Marakwet

80 Poverty Gap Ranking 70 60 50 40 30 20 10 0 Kisii Kilifi Kitui Nyeri Meru Siaya Busia Embu Nandi Kwale Vihiga Migori Bomet Nairobi Nakuru Kisumu Kericho Kiambu Baringo Laikipia Nyamira Makueni Lamu (4) Kirinyaga Murang'a Bungoma Mombasa Wajir (44) Wajir Homa Bay Isiolo (42) Machakos Kakamega Nyandarua Narok (10) Kajiado (1) Trans Nzoia Trans Taita Taveta Taita Uasin Gishu Garissa (26) Turkana (47) Turkana Tharaka Nithi Marsabit (45) Mandera (46) Samburu (43) Tana River (41) Tana West Pokot (37) Pokot West National Average Elgeyo/Marakwet

Source: KIHBS 2005.

financing for a fairer , more prosperous kenya 19 one-third of the poverty threshold. At the other end nificance in Kenya, not least because of the strong of the scale, Kajiado has the lowest level of household correlation between household wealth and indicators income poverty in the country; with Narok and Lamu such as school attendance (see below), child survival also well below the national average. and nutrition.

Integrating the poverty gap into the formula for inter- Among the many caveats that have to be attached to governmental transfers requires a disaggregation of income poverty and wealth distribution data for the 12 the national gap into county shares. Taken individu- counties, two related concerns merit specific mention. ally and collectively, the 12 ASAL counties account for First, the 2005 KIHBS provided a static (and by now a larger share of the national poverty gap than their dated) snapshot of household consumption at one population share (Figure 2). The outlier is Turkana, point in time. The state of poverty itself is dynamic, which has a poverty gap some four times larger than with populations moving above and below the poverty its population share. Other counties such as Marsabit, threshold over different periods. West Pokot, Isiolo and Wajir also account for a share of the national poverty gap far exceeding their population Second, the arid and semi-arid areas of Kenya are shares, while the inverse holds for Narok and Kajiado. It characterized by low and erratic rainfall and highly should be noted that the ASAL counties are not alone vulnerable livelihoods. In this context, income-based in having an oversized share of the national poverty indicators can provide at best a very partial indicator gap, as witnessed by the data for counties such as of the risks and vulnerabilities that come with , Baringo, Kakamega and Machakos (see Figure 2). loss of livestock and food insecurity. Pastoralists have developed sophisticated coping mechanisms to man- age risk. These include moving herds and social insur- Household Consumption: High Levels of ance arrangements, such as the transfer of breeding Inequality animals. Even in a normal year these arrangements Kenya’s county-level poverty profile mirrors some come under pressure (Fitzgibbon 2012). During pe- deep horizontal disparities in the distribution of in- riods of severe drought these arrangements break come. People living in the ASAL counties are con- down in the face of rising food costs and falling prices centrated in the lower reaches of Kenya’s income for livestock, and the depletion of herds. The devas- distribution. tating drought of 2010 and 2011 left some 3.7 million people facing chronic food insecurity in seven coun- The data are striking. Over 50 percent of households ties (Turkana, Mandera, Marsabit, Garisaa, Wajir, Isiolo in Samburu and Mandera are in the poorest quintile and Tana River) with among the highest incidence of Kenyan society, rising to 86 percent for Turkana. of poverty in Kenya. The combination of rising food To view the data from the other end of the wealth prices—food price inflation stood at 11 percent in 2011 telescope, a child born in Turkana or Wajir has a 1-2 —and declining prices for livestock will have pulled a percent chance of being born into the wealthiest significant number of people below the poverty line, quintile. The equivalent figure for Machakos is 21 per- pushed many of those already in poverty further be- cent, rising to 75 percent in Nairobi (Figure 3). These low the poverty threshold, and contributed to acute county-level wealth disparities are of enormous sig- nutritional problems (World Bank 2011a).

20 Global Economy and Development Program Figure 2: Poverty and Population: County Shares of National Poverty Gap and Population

7 Turkana


5 Kakamega Machakos 4

3 Share of National Poverty GapShare of National (percent) Poverty

Wajir Baringo Mandera 2 West Pokot 0 Marsabit Tana River Narok 1 Samburu Garissa Isiolo Kajiado 0 Lamu 1 2 3 4 5

Share of National Population (percent) Source: Census 2009 & KIHBS 2005. * Nairobi is not included in the data. The county accounts for 7.9 percent of the population and 3.8 percent of the poverty gap.

Health and Nutrition counties. In this section we draw on the DHS and wider health survey data disaggregated to follow the con- Using the Millennium Development Goals targets as tours of the new counties. a benchmark for measuring progress, Kenya has a mixed record on health and nutrition indicators. There have been remarkable gains on some indicators – and The National Picture little progress on others. As in other countries, health The 2008-2009 DHS records a number of major ad- status in Kenya is the result of many important factors vances in public health. One of the most positive find- including the provision of basic services and health ings to emerge was a sharp decline in child mortality. inequalities linked to gender, geography and socio- Between 2003 and 2008, the under-five death rate economic status. Data constraints make it impossible declined from 115 to 74 deaths for every 1000 live to document trends in the ASAL counties relative to births—a 36 percent drop. The record on the nutri- the rest of Kenya, although the data that is available tional status of children has been less encouraging. highlights some distinctive challenges. The 2008- There was a modest decline in stunting between 2000 2009 Demographic and Health Survey (DHS) provides and 2008—from 35 percent to 30 percent—with re- the most recent overview of the health status of ported stunting increasing in North Eastern Province.6 Kenya’s people, however the survey data is organized The proportion of children who are wasted and under- on the basis of the old provinces rather than the new weight changed little in the decade after 2000, raising

financing for a fairer , more prosperous kenya 21 Figure 3: For Richer, for Poorer: Share of Population concerns over the degree to which in Top and Bottom Quintile of the Wealth Distribution increased income has translated into (47 counties) reduced poverty.

Turkana The nutritional status of children Mandera Samburu should weigh heavily in any consid- Marsabit eration of equity in public spending. Wajir Tana River Apart from the immediate concerns Isiolo over humanitarian suffering, malnu- Kilifi trition in the early years sets children Makueni West Pokot on course for a life of disadvantage, Kwale vulnerability and underachievement. Busia Kitui Those affected are less likely to enter Kisii school at an appropriate age and less Tharaka Nithi Baringo likely to make the transition to second- Nyandarua ary school. Moreover, there is com- Machakos pelling evidence that early childhood Bungoma Kakamega malnutrition inflicts damage on cogni- Elgeyo/Marakwet tive development. As a recent series in Migori Embu The Lancet powerfully documents, the Nyeri combined effects of household poverty Homa Bay and poor nutrition affect brain devel- Kisumu Taita Taveta opment from the prenatal period or Garissa earlier (The Lancet 2007; The Lancet Narok Trans Nzoia 2008). That damage, which is often ir- Vihiga reversible, is reflected in lower levels of Nandi Uasin Gishu education attainment and lower levels Laikipia of income (The Lancet 2011). Bomet Nakuru Kericho Progress on maternal mortality, the fifth Nyamira of the MDGs, is uncertain. The 2008- Siaya Murang'a 2009 DHS reported a small increase in Kiambu the maternal mortality rate, while up- Meru dated estimates prepared on the basis Kirinyaga Kajiado of a more recent tracking survey points Mombasa to a sharp decline (Hogan et al 2010). Lamu Nairobi Divergent estimates point to the large margins of error in sampling. To the ex- 100 80 60 40 20 0 20 40 60 80 100 tent that any definitive conclusion can Source: KIHBS 2005.

22 Global Economy and Development Program be drawn, maternal mortality remains high in Kenya. is the relationship between stunting and underweight The maternal mortality ratio is 488 deaths for every prevalence. However, several of the 12 ASAL coun- 100,000 live births. Risks are associated with differen- ties register particularly worrisome levels of depri- tial levels of wealth, education, birth-spacing, access to vation. They account for six of the 10 counties with health facilities and other factors. The 2008-2009 DHS the highest prevalence of underweight children. In reported North Eastern Province as having the lowest five of these – Turkana, Tana River, Mandera, Isiolo proportion of births delivered in a health facility—just and Samburu –more than one child in every three is 17 percent compared to 89 percent in Nairobi (Kenya underweight for their age (Figure 4). Extreme stunt- National Bureau of Statistics 2010). When questioned ing levels (three standard deviations or more from as to reasons for not delivering in a health facility, 17 the predicted height-for-age) provide an indicator of percent of mothers in the North Eastern Province iden- sustained and chronic nutritional deprivation. Four of tified the poor quality of service available as the major the ASAL counties register particularly high levels of concern (four times higher than in any other district) extreme stunting with over 25 percent in Garissa and and 9 percent cited the fact that there was no female 40 percent in Wajir affected. provider (no other province registered this as a concern for more than 1 percent of women). The proportion of As in the case of income poverty, there are dan - women in the North Eastern Province citing distance as gers in reliance on static snapshots of malnutrition. a barrier to delivery in a health facility was also the high- This is especially true for the ASAL counties, where est in Kenya. nutritional status—particularly proportions of un- derweight children—varies significantly within and between seasons, and over time. Additionally, drought The 12 Counties can have dramatic effects on nutrition that may not Data availability does not allow for cross county com- be captured by occasional surveys. Research carried parisons of some key indicators, including child and out by Save the Children during the 2011 drought in maternal mortality. Drawing on the 2005 household Wajir and Mandera found global acute malnutrition consumption survey it is possible to derive a picture rates of 23 percent and 32 percent respectively (the – albeit somewhat dated and partial – of nutritional in- World Health Organization’s emergency threshold is dicators, and of access to basic health care. Data from 15 percent). In both cases, the levels registered were the Health Management Information System provides some four to five times the rates documented in 2009. another data source allowing for disaggregation to Findings such as these illustrate the degree to which the county level in some areas of service delivery. pastoralist households with limited savings and high levels of poverty are ill-equipped to cope with the Nutritional Indicators combined effects of rising food prices and declining The national picture on child malnutrition is disturb- livestock prices. ing across Kenya. Many counties with relatively high average income levels and a low incidence of income Access to Basic Services: Immunization and poverty have a high incidence of stunting and under- Birth Attendance weight children. The relationship between income and Access to health facilities and the availability of skilled nutritional status is decidedly nonlinear in Kenya, as staff are two of the most critical factors influencing

financing for a fairer , more prosperous kenya 23 Figure 4: The Nutritional Status of Kenya’s Children: Extreme Stunting and Underweight- for-Age (47 counties)

45 Wajir 40


30 Garissa 25 West Pokot Tana River 20 Mandera Turkana Samburu 15 Marsabit Lamu Narok Stunting: 3D (percent) Stunting: 10 Kajiado Isiolo 5

0 0 5 10 15 20 25 30 35 40 45 50

Underweight: 2D (percent)

Source: KIHBS 2005. *Stunting 3D: Children whose height-for-age is below three standard deviations from the mean are said to be severely stunted. *Underweight 2D: Children whose weight-for-age is two standard deviations from the mean are said to be moderately underweight. *Lamu and Narok are overlapping in the figure.

opportunities for health. To varying degrees, the The consequences of these disparities in access to ASAL counties covered in this survey lose out on both qualified medical care are apparent in a range of fronts. health indicators. While immunization rates have gen- erally improved in recent years, all but three of the Data provided by the Commission on Revenue 12 counties are in the bottom half of the league table Allocation in 2012 highlights the extent of national for full vaccination coverage. The limited presence of inequalities (Table 2). On average across Kenya’s 47 health providers is reflected in the high proportion of counties there is one doctor for every 25,000 people, births not attended by skilled medical staff. Five of and one nurse for every 2,054 people. Almost all the seven counties with the lowest rates of coverage of the 12 counties have ratios above both levels. In are in the bottom seven of the national league table Turkana, the ratio of people-to-doctors is more than on this indicator. In both Turkana and Wajir, only 5-6 10 times the national average and the ratio of people- percent of births are attended by skilled providers, to-nurses is seven times the national average. which is less than one-third of the national average.

24 Global Economy and Development Program Inequalities such as these can only be addressed perceptions of the quality of staff and service provi- through public spending measures that allocate re- sion for birth attendance. sources against need. Significant barriers to access persist. Cost and dis- As in other areas, the health service delivery picture tance have a marked bearing on access to health ser- is not straightforward. Some of the 12 focus counties vices across Kenya—and the 12 ASAL counties are no are near the bottom of the national ranking for both exception. Over one-third of total health spending in immunization and skilled birth attendance— Wajir, Kenya takes the form of out-of-pocket payments. While Mandera and Turkana are examples. Others appear this share has been shrinking with the rise in public close to the top of the national ranking on one indica- spending, cost remains a substantial obstacle for poor tor, but closer to the bottom on another—West Pokot Kenyans—and the high levels of poverty in the ASAL and Isiolo do far better on immunization than birth counties raises the height of that barrier. Distance is attendance, and vice versa for Lamu. These outcomes another barrier. In some of the larger ASAL counties it illustrate the differential effects of government pro- is not uncommon for communities to be located more grams and priorities. They may also reflect public than 30 kilometers from the nearest health facility.

Table 2: Immunization and Qualified Medical Assistance at Birth: Ranking of 12 Counties and National Average Medical Fully assistance Rank Vaccinated Rank Population Rank [out during birth [out of 47 [children under [out of 47 Per Doctor of 45 County (percent) counties] five] (percent) counties] (in 000’s) counties] Garissa 23.9 34 74.6 25 52 29 Isiolo 27.9 29 72.2 30 143 39 Kajiado 39.8 18 70.7 31 76 34 Lamu 27.2 30 80.5 19 No data n/a Mandera 11.3 45 47.0 46 256 41 Marsabit 17.4 41 80.1 20 32 18 Narok 18.9 40 62.2 42 41 22 Samburu 19.0 39 85.6 13 No data n/a Tana River 20.4 38 85.7 12 48 28 Turkana 6.9 46 66.7 35 285 44 Wajir 5.4 47 72.7 28 132 38 West Pokot 16.9 42 56.2 43 73 33 National 37.6 - 75.0 - 25 - Average Source: Commission on Revenue Allocation 2011.

financing for a fairer , more prosperous kenya 25 Even when able to reach a facility, there is no guar- in marginalized areas in order to meet the level of antee that patients will receive effective treatment. quality achieved in the rest of the nation. Against this Further, national surveys have pointed to acute short- backdrop, there would appear to be strong grounds ages in both staff and medicines in health facilities for ensuring that devolved financing in health in- across ASAL districts (Government of Kenya 2010). cludes special provisions for those ASAL counties facing acute shortages of facilities, trained health The new constitution identifies public spending as a workers and medicines. means to enhance the provision of health facilities

26 Global Economy and Development Program Education: Access and inequalities in opportunities for education. This Learning applies both to school participation and learning achievement. The ASAL counties represent areas of ducation has been a partial success story in acute deprivation, with restricted opportunities for EKenya over the past decade. Enrollment rates education reinforcing and interacting with wider so- have increased at all levels. More children are enter- cial disadvantages. More equitable patterns of public ing primary school, completing the primary cycle, and spending harnessed to more effective policies for de- making the transition to secondary school. On a less livering quality education in marginalized areas could positive note, the Kenyan education system is charac- play a decisive role in unlocking the potential of edu- terized by continued problems in access, high levels cation as a catalyst for accelerated growth, poverty of inequality and low levels of learning achievement. reduction and human development.

Tackling the twin challenge of unequal access and poor quality provision is central to the realization of The National Picture: School the ambition of transforming Kenya into a dynamic, Participation and the Quality of inclusive, middle-income country, as articulated in the Education Vision 2030 strategy. Addressing these two key chal- The Kenyan government has placed considerable em- lenges is also essential to sustained progress across a phasis on increasing access to education. In 2003 a wider range of human development indicators. Higher policy of Free Primary Education (FPE) was adopted, levels of education, especially maternal education, are leading to the withdrawal of formal fees for primary inversely correlated with child death rates and malnu- school. More recently in 2008, a policy of ‘free sec- trition, and positively correlated with the use of basic ondary education’ was introduced in an effort to services (Table 3). ensure that children from poor households acquire a quality education that enables them to access op- Greater equity is critical if Kenya is to unlock the portunities for self-advancement. potential of education as a force for change. The country is marked by extreme vertical and horizontal

Table 3: Maternal Education and Wider Development Indicators

Development Indicator Mothers Under-five Skilled antenatal Delivered by skilled Extreme Child Education mortality care provider Stunting (per 1000) (percentage) (percentage) (percentage) None 86 72 19 17 Primary 68 95 49 14 Secondary or 59 96 73 9 Higher Source: DHS 2008.

financing for a fairer , more prosperous kenya 27 School Participation: A Rising Tide of 5 to 6 years for males in the same time period (Kenya Enrollment National Bureau of Statistics 2010). Measured by headcount numbers the effort to acceler- ate progress towards universal primary education has The surge in enrollment since 2000 has brought large delivered results. Between 2002 and 2009, the number numbers of over-age children into the education sys- of children enrolled in primary school increased from 6 tem. Many of these children would previously have million to 9.5 million. The net enrollment rate increased been excluded from school by the cost of education. from 79.9 percent to 92 percent over the same period Others would have been re-entering the system hav- (Figure 5). Part of the surge in enrollment has been ing previously dropped out. Over 70 percent of the absorbed by private schools, although public provision children in Kenya’s primary school classrooms are still overwhelmingly dominates the education delivery older than the prescribed age for their grade. The landscape, accounting for 88 percent of enrollment at national age-for-grade profile is captured in Figure 6, the primary level in 2008 (Government of Kenya 2011b). which documents the presence of almost a half mil- An additional 1 million children entered the secondary lion 8-10-year-olds in Standard 1, and 150,000 children system between 2002 and 2009, with gross enroll- aged 13-year-olds in Standard 5 (two years over the ment rates rising from 29 percent to 42 percent. The prescribed age-for-grade). This profile has some im- overall gains have seen the median number of years of portant implications for the quality of education and schooling completed (for those aged over 6 years) rise the additional challenges associated with teaching from 4.3 in 2003 to 5.2 in 2008 for females and from over-age children.

Figure 5: Progress in Education: Male and Female Enrollment Rate (2000-2009)


Sec GER Female 100

Sec GER Male 80

Prim NER 60 Prim GER Female 40

Enrollment Rate (percent) Prim GER Male


0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Year

Source: EMIS 2000-2009.

28 Global Economy and Development Program Figure 6: The Age Profile in Kenya’s Classrooms: Age-By-Grade Enrollment (2010)

500 250,000 children start school 150,000 13 year olds 450 at age 8 are in Standard 5 400 350 Std 1 300 Std 2 250 Std 3 200 Std 4 Enrollment (000) 150 Std 5 Std 6 100 Std 7 50 Std 8 0 below 6 6 7 8 9 10 11 12 13 14 15 Age Source: EMIS 2010.

Gender disparities have proven resilient to change in That distance is reflected in out-of-school numbers. As both primary and secondary education, where they illustrated by the data in Table 4, any estimate of out-of- have increased since 2006 (Figure 5). By Standard 8, school numbers in Kenya is subject to large margins of there are just nine girls in school for every 10 boys. While error related to divergent estimates for the denominator girls have a slightly higher transition rate from primary (the number of children) and the use of different indica- to secondary school, the gross secondary enrollment tors for the nominator (the number of children enrolled rate for girls is 42 percent and 49 percent for boys—a or attending school). Estimates by the UNESCO Institute discrepancy that is equivalent to around 108,000 ‘miss- for Statistics (UIS), the primary international reporting ing girls.’7 Age-specific school attendance rates point agency, put the out-of-school number at around 1 mil- to higher levels of attendance by males at ages 5-6, lion for 2009. This is above the government of Kenya’s reflecting the delayed entry of girls into basic educa- own estimate (around 600,000) based on enrollment tion. Gender disparities in attendance equalize around data in the Education Management Information System age 13-14, before widening in favor of males from age (EMIS), which reports a higher enrollment rate than 14 onwards (Kenya National Bureau of Statistics 2010). that used by the UIS. However, neither of these sources Significant gaps in wealth cut across the gender dispari- uses the 2009 population census, which revised up- ties, especially at the secondary school level (Figure 7). wards estimates for the size of the school population.8 Applying the net enrollment rate reported in the EMIS to census population for primary school-age children Out-of-School Numbers would put the number of the out-of-school children to Headline figures on national enrollment have to be around 2 million. Survey data on school attendance, as interpreted with some caution. While Kenya is getting distinct from administrative data on enrollment, tells a more children into the school system, there are sig- different story again. The 2009 census and the 2008- nificant gaps and problems with retention. If the mea- 2009 DHS report school attendance rates of 77 percent sure of universal basic education is the proportion of and 79 percent respectively, implying an out-of-school children progressing through the full national cycle of population of around 1.9 million in the case of the eight years, Kenya still has some distance to travel. census.

financing for a fairer , more prosperous kenya 29 Figure 7: Kenya’s Wealth Gaps in School Attendance (2008) 100




60 National Average 50 Richest 40 Poorest School attendance (percent) 30



0 Primary (6-13 years old) Secondary (14-17 years old)

Source: DHS Report 2008-2009.

Table 4: Reported School Attendance and approach to public spending. If is a Enrollment basic constitutional right, then the government needs credible and robust indicators to assess the number Attendance/ Out-of-school of children denied that right – and to estimate the fi- Enrollment estimate nancing requirements for delivering education for all. Source (percentage) (millions) Second, if educational disadvantage is to be included Census 2009* 77 1.9 as an element in national financing formulae, then * DHS 2008 79 1.8 an accurate county-level profile of school participa- Uwezo 2011* 87 1.2 tion is a required guide for resource allocation. Third, National there are worrying signs that, whichever baseline is Administrative Data** 90 1.1 used, Kenya is struggling to maintain the momentum *Out-of-school population calculated using the Census 2009 towards universal net enrollment. One reason for this Primary School Population (ages 6-13). is that, like other countries, Kenya now faces the chal- ** As reported on the Global Monitoring Report 2011 (ages 6-11). lenge of extending opportunities to children who are the hardest to reach – the last 10-15 percent of the National data on out-of-school numbers raise three primary school-age group in this case. Many of these related sets of concerns. First, the discrepancies children live in the ASAL counties. in the data have far reaching implications for any

30 Global Economy and Development Program School Progression in 2010. However, there were just 740,000 students As the out-of-school numbers indicate, progression in Standard 8 in that year, and fewer than 400,000— through Kenya’s education system remains difficult or one-third of the 2003 intake number— took the for many children. There are high levels of attrition Kenyan Certificate of Primary Education (KCPE) at the at various points in the school cycle, including in the end of the primary cycle. early grades into the last grade of the basic education cycle, and in transition to secondary school. These figures suggest that repetition and drop out take a heavy toll. Many children do not make it

Pure cohort tracking is not possible in Kenya because through the basic education system in the anticipated the absence of longitudinal data makes it impos - number of years and many do not make it through the sible to track identifiable children. Some indication of system at all. There is further attrition at the second- progression patterns can be created through proxy ary school level. In 2010, a reported 354,000 students tracking exercises, which trace classroom numbers sat for the Kenyan Certificate of Secondary Education across grades. Figure 8 summarizes such an exercise (KCSE), implying that 10 percent of entrants to sec- for the 1.3 million children who entered Standard 1 in ondary school in 2007 had either dropped out or not 2003. With smooth progression, these children would yet completed Standard 4. have been expected to complete an eight year cycle

Figure 8: Charting Grade Progression: Reported Enrollment for Standard 1 Through the KCSE (2003)

1400 Total 1200 Girls Boys 1000

800 Enrollment 600



0 2003 2004 2005 2006 2007 2008 2009 2010 2010* Std 1 Std 2 Std 3 Std 4 Std 5 Std 6 Std 7 Std 8 Year and Grade Source: EMIS 2010. *Number of students sitting for KSCE.

financing for a fairer , more prosperous kenya 31 Factors Keeping Children Out of School where the education of girls is perceived as being and Fueling Attrition of less value than the education of boys, economic pressures are likely to fuel gender disparities. There has been extensive research into the barri- ers that keep children out of school and the factors • Education quality: Parents in poor households behind school attrition. While there are many local- have to make considerable investments to put their ized variations, five major and overlapping themes children through school. To the extent that these investments are perceived to generate returns in emerge: terms of improved prospects for employment and • Parental education: As in other countries, in Kenya wider opportunities, parents will have an incentive participation in education is strongly associated to keep their children enrolled in school. However, with parental education. Having literate parents those incentives will be weakened if the education confers significant advantages that may be associ- system is seen as delivering limited results—and ated with the value attached to education, support some of the learning achievement results are dis- with homework, parental confidence in engaging couraging (see below). In some areas, including the with schools and teachers, and household wealth ef- ASAL counties, parental concerns over quality ex- fects. Disaggregated county-level data on parental tend to the school curriculum itself, which may not education is not yet available. However, the 2008- be seen as sufficiently sensitive to local language, 2009 Demographic and Health Survey found that beliefs and customs, or as sufficiently relevant to over two-thirds of women (and one-half of men) in livelihoods. the old North Eastern Province reported no educa- • Health effects: The poverty and childhood health tion, compared to just 6 percent in Nairobi and 10 and nutrition indicators discussed earlier in this re- percent in . port have far reaching consequences for education. • Household wealth: Cost remains a major barrier Cross country research has demonstrated that both to education for the poorest households in Kenya. stunting and poverty are associated with reduced Despite the policy of Free Primary Education, par- years of schooling and lower test scores. One of ents still face indirect costs including uniforms, the most detailed studies finds that being stunted learning materials and a range of informal charges and living in poverty results in a loss of two years (even though FPE funds include support for learn- of schooling and another two years of lower grade ing materials). Under the 2008 policy, the govern- attainment (Martorell et al 2010). The cumulative ment of Kenya has committed to providing a per effects of illness and micronutrient deficiency in pupil subsidy for all children in public day second- terms of lost school attendance, diminished cogni- ary schools, but schools still charge to cover the tive development and lower learning outcomes has costs of development projects and food. As such, not been estimated for Kenya—but the costs are it would be more accurate to describe the policy likely to be very high. as one of reducing charges.9 One study estimates • Distance and gender: Low population density that these costs amount to as much as $186 a year and fewer schools in some of the 12 counties result for day school pupils and $368 a year (at 2007 in longer distances and journey times to school, exchange rates) for public boarding school pupils which are in turn associated with lower attendance (Obha 2009). Household expenditure for secondary rates for children who are not in boarding school. school averages eight times the level for primary Problems are compounded at the secondary level education (Glennerster et al 2011). For the reported because there are fewer schools. In counties where 45 percent of Kenyans living below the poverty adolescent children are actively engaged in herd- line, these are significant cost barriers. In a setting ing and in water and firewood collection, distance

32 Global Economy and Development Program to school is associated with high opportunity costs. While the Uwezo surveys attract considerable media Gender factors also come into play, with parental interest in Kenya, the implications of the results are security fears militating against allowing girls to not sufficiently recognized. Consider the test results walk long distances or to join boarding schools for final grade students on the Standard 2 division (UNESCO 2010). sum. Fully 10 percent of these students have gone through seven years of schooling with no value added The National Learning Achievement in terms of their ability to perform foundational skill Deficit tasks from Grade 2. Other sources broadly corrobo- The primary focus for basic education policy in Kenya rate the Uwezo survey results. In 2010, results from a over the past decade has been getting children into survey carried out by the Kenya National Assessment school. Less attention has been directed to what chil- Center found that half of pupils in Standard 6 were dren learn in school. As in many other countries, the unable to achieve basic competency levels for literacy emerging evidence strongly suggests that a more and numeracy (Wasanga, Ogle and Wambua 2010). integrated approach is needed. The next generation of reform needs to combine an equal commitment to These learning shortfalls inevitably contribute to the enhanced access and learning. high rates of attrition recorded in the previous sec- tion. Children who aren’t learning are less likely to progress across grades, in part because their parents Uwezo Surveys: ‘Our Children Are Not may be unwilling to meet the direct financial costs and Learning’ wider opportunity costs of keeping them in school. It Surveys carried out by Uwezo, a Kenyan nongovern- can reasonably be assumed that a large proportion mental organization, have highlighted the poor state of the dropout that occurs between Standard 7 and of learning in many of Kenya’s schools. These surveys Standard 8 is a direct result of parental recognition test children in higher grades on exercises designed that their children are unlikely to pass the school- for lower classes. Apart from documenting the abso- leaving test. lute level of learning, they capture the value added by a year of education as children progress across grades (Uwezo Kenya 2011). National Examination Results: Primary School

The results tell their own story. The 2011 Uwezo survey Examination results provide another window into revealed that some 70 percent of children in Standard Kenya’s learning achievement problems. Many of 3 were unable to successfully complete tests designed Kenya’s children fall far short of the learning achieve- for Standard 2 children. More alarmingly, one in five ment levels required to progress to secondary school. children in Standard 4 could not read a text designed This is true even for those who progress through for Standard 2 children; and 9 percent of children in primary school to sit for the Kenya Certificate of Standard 8, the final grade of primary school, could Primary Education, the results of which are used to not do a Standard 2 division sum. As the survey con- select children for entry to one of the three secondary cluded, “Our children are going to school, but they are school tiers—district, provincial and the elite national not learning.” schools.

financing for a fairer , more prosperous kenya 33 Some caution has to be exercised in using test scores test score distribution, scores below 200 are consid- to assess learning achievement levels over time. The ered very poor and well below the level required to scores are normalized with the distribution pattern make a successful transition to secondary education. geared in part towards the availability of secondary Students scoring at this level are registering learning school places. Even so, a large proportion of students achievement standards several grades below those score at a level so low as to raise questions about the expected. The range between 200 and 250 is also con- value added by eight years of schooling. sidered to be below the required level for secondary schooling, with students scoring in this range requir- Figure 9 documents the test score distribution for ing remedial teaching to make up lost ground. As indi- 2010. It identifies a number of performance bands cated by the test score distribution, fully one-quarter for the five core competencies covered by the KCPE of KCPE students, 171,000 in total, score below 200. (Kiswahili, English, math, science and social studies). On the other side of the low performance threshold, While there is no ‘pass’ or ‘fail’ cut-off point in the 28 percent of students score between 200 and 250.

Figure 9: Kenya’s Primary School Learning Outcome Results: National Frequency Distribution for Test Scores (2010)

-3SD -2SD -1SD Mean +1SD +2SD +3SD 250 24 percent of students scored 210000 less than 200 – 190000 200 below basic competency

150 130000

100000 100 Frequency (Thousands) Frequency 54000 50 31000 11000 30 1251 0 0 50 100 150 200 250 300 350 400 450 500 Score Source: KNEC: KCPE 2010.

34 Global Economy and Development Program In summary, just over half of the students making it Private School Attendance: A through to the KCPE exam stage are scoring at lev- Significant Advantage els below those required to prepare for a successful Attendance at a private primary school confers signifi- secondary education. For the 15 percent of students cant advantages for KCPE candidates. Private school scoring below 185, the performance level is so low students score higher grades and are more likely to be as to raise questions about the value added by their eligible for entry to the elite tier of national schools. schooling over a period of several years. Paradoxically, a private primary education is the most secure route into a high quality, publicly financed sec- Looking beyond primary school the number of stu- ondary education. dents sitting the Kenya Certificate of Secondary

Education has been rising steadily. In 2010, a record The learning achievement advantages registered by 354,062 pupils sat the exam. Just over one-half of private school students are apparent long before they these pupils scored a C+ or higher. Reflecting the sit the KCPE. The 2010 National Assessment Center gender gap in secondary school enrollment, boys ac- survey found that pupils in private schools were counted for 55 percent of candidates. Gaps in test outperforming their counterparts in public school score appear to rise with the grade. Thus while an by around two-thirds of a standard deviation on nu- approximately equivalent share of male and female meracy and one standard deviation on literacy by candidates gain a C+ or higher, boys are more likely to Grade 5. By the time children sit the KCPE the perfor- gain a B+ or higher. Nationally, 6 percent of Standard mance gap is extremely large. In 2011, there were just 4 leavers were admitted to university in 2009, with two government schools in the top 30 of the national a further 14 percent admitted to technical and voca- KCPE ranking; and just 10 in the top 130. While repre- tional courses or college. senting only around 10-15 percent of KCPE candidates, private primary school graduates typically account for There are marked disparities in performance across around 50 percent of the KCSE candidates in national schools. Data for national schools in 2008 indicate schools (Glennerster et al 2011). that 90 percent of students scored a C+ or higher, with an average score of 9.6 out of 12. Gender gaps Some commentators point to test score differences as in score were insignificant. In provincial schools the evidence of the inherent advantages of private over share of students scoring C+ dropped to 43 percent, public schools. That evidence is in turn cited to make with significant gender gaps emerging. In district the case for expanding the provision of public finance schools just 11 percent of students scored C+ and the for low-fee private schools, notably through vouchers proportion of boys performing at this level was almost for children from low-income households. Does the double that for girls (Glennerster et al 2011). While the undoubted public-private school performance gap government of Kenya has recently moved to imple- justify the policy prescriptions in favor of publicly fi- ment quotas for public school entrants to national nanced private education? and provincial schools, these test disparities serve to underline the advantages that come with attendance Not on the basis of the evidence presented to date at the high-performing private schools at the primary (Box 2). While several studies have sought to dem- level (Muindi 2012a; Muindi 2012b). onstrate that private schools outperform public

financing for a fairer , more prosperous kenya 35 schools, most have failed to adequately control for That observation almost certainly has a far wider the socio-economic status of pupils. They have also application. It draws attention to problems of the failed to sort children by the type of private schools regulatory failure of nonstate providers in a context of they attend. The private school sector in Kenya is widespread state failure to deliver good quality edu- very diverse, spanning 10,000 registered schools and cation. Even established private school associations an unknown number of unregistered schools. Some have expressed concern that the rapid growth of the of these schools, especially the best performing sector has seen an increase in malpractice—ranging among them, are drawing pupils from high-income from the corrupt purchase of exam papers, the reg- households. At the other end of the spectrum are istration of weak students under the names of other low-fee private schools operating in informal urban schools, and the practice of poaching top students settlements and some rural areas. There are no robust from public schools—to drive up results. The Kenya studies comparing the low-fee schools serving poor Private Schools Association has called on the govern- communities with public schools serving comparable ment to introduce legislation requiring more stringent communities. regulation.

None of this is to understate the scale of the learn- Factors Behind Low Levels of Learning ing achievement crisis in Kenya’s public schools. Achievement State failure to deliver quality education has fuelled a large-scale exit from the public school system. As in the case of school access and retention, many There are now over one million children in private of the barriers to improved learning achievement are primary schools—some 10 times the number before well understood even though more research is needed the introduction of free primary education in 2003 to identify strategies for raising standards. (Government of Kenya 2008). Many of these chil- dren come from exceptionally poor households, and Some of the factors holding down learning standards those in low-fee private schools are often paying for are exogenous to the education system. Poverty, mal- an education of exceptionally poor quality. Indeed nutrition and parental illiteracy clearly disadvantage concerns over the standards of low-fee providers many children. Early childhood provision has the po- have increased with their expansion. During 2011 the tential to mitigate that disadvantage. However, only District Education Board in Kisii ordered the closure one-quarter of children in the relevant age group of 30 low-fee private academies, 20 of which were are enrolled in pre-primary education and there are among the worst performers in the national KCPE marked disparities across income groups and regions. ranking (Nyagesiba 2012). The district commissioner’s report highlighted the failure of the schools to comply School-based and wider institutional failings in the edu- with the basic standards for environmental safety and cation system hamper learning prospects. Shortages learning set by the Ministry of Education. “The owners of textbooks and teaching materials are a problem, es- of the schools,” he commented, “are interested only pecially for children from households unable to afford in making money at the expense of young learners.” them. The 2011 Uwezo survey found just one textbook to (Nyagesiba 2012) every three children in Class 2. This is broadly consistent

36 Global Economy and Development Program Box 2: Low-fee Private Schools in Kenya: Symptom of Between 2003 and 2008 both the public and private State Failure, or Cure for the National Learning Crisis? school sector registered enrollment increases in excess of 700,000 pupils, with private schools enrollment increas- Since the nationwide elimination of school fees in 2003, in ing from 2.6 percent to 10 percent. The headline figures Kenya the private school sector as a whole has grown rapidly. do not capture underlying patterns of school and pupil Within this sector, many of Kenya’s poorest households, es- segmentation. Given that the rise in enrollment was as- pecially those living in informal urban settlements, send their sociated with the lowering of cost barriers, it is probable children to low-fee private schools. Several commentators that most of the 1.4 million pupils entering the education have argued that these schools represent a viable, afford- system for the first time were from the poorest, least able and cost-effective alternative to poor quality public pro- literate households in Kenya. Many would have been vision—and that both the Kenyan government and donors first generation learners. Meanwhile, most pupils exiting should be using school vouchers and other arrangements to public schools and entering the private system were, by increase budget financing for private education. definition, able to afford the transition. In other words, the marginal student entering the public system was carrying Are these claims and policy conclusions supported by rigor- a higher level of educational disadvantage than the mar- ous evidence? The underlying arguments rest on the conten- ginal student entering private schools. The 90 percent of tion that low-fee private schools are delivering higher levels pupils attending Kenya’s public schools include the most of learning achievement at lower cost than public schools— deprived in the country, while private schools include the and that provision could be scaled up without compromising most advantaged (including post-2003 recruits from pub- the quality and cost advantage. Current research evidence lic schools). does not support these conclusions. Moreover, while many poor households have exited public schools because of qual- • Failure to sort by pupil and school identification. In 2005 ity concerns, advocates for low-fee private schools have (the last data point for the survey) most low-fee private tended to neglect the absence of state provision as a ‘push schools were unregistered. Their pupils took their KCPE factor’. exams in public schools—and their results were recorded as public school results. The practice remains widespread One recent study illustrates some of the weaknesses under- even today, yet the survey does not control for the con- pinning the case for an increase in public finance for private sequences of this important administrative practice. The education (Bold et al 2011a). In a review of KCPE data up to sorting problems do not end here. Most low-fee private 2005, the authors find that private schools achieved an aver- schools operating in informal settlements provide classes age test score premium of around 20 percent—equivalent to only up to Form 6 or below, while the KCPE exam is taken one full standard deviation. With reported average per pupil by Form 8 students. It therefore appears likely that many cost in two-thirds of private schools (as measured by re- low-fee private school pupils either drop out or transfer to ported school fees paid by parents) being less than half pub- public schools, making it difficult to attribute achievement lic spending per pupil in public schools, the authors contend gains by school type. that increased public spending on private schools could raise • Failure to control for dispersion of private school funding education quality at a net savings to the national budget. and household finance. As in other countries, the private school sector in Kenya is very diverse. The authors of The disarmingly simple policy conclusion obscures some the study under review estimated the 2006, median and serious methodological flaws. Among the problems with the mean private school fees per pupil respectively at $40 research and the subsequent policy conclusions it draws, and $110 per year. They contrast this with an estimated others include: average per pupil cost of $88 per year in state schools. However, the data comparison does not include non-fee • Failure to control for the socio-economic status of pu - expenditures undertaken by households of children in pils. The superior performance of private schools in the private schools—a major omission in any comparison of KCPE exams is well documented (see main text). However, cost effectiveness. Moreover, it is not clear that the survey matching pupils to compare like-with-like is difficult covers the bulk of students attending the low-fee private in Kenya—and the study fails to address the problem. school sector. The 2005 Kenya Integrated Household

financing for a fairer , more prosperous kenya 37 Budget survey reported that 47 percent of the country • The emergence of the low-fee private school in Kenya is a had a income of $38 per month or less. The implication is response to various underlying currents. Concern over the that the cost of sending two children to the median low- quality of public provision is certainly one of those cur- fee private school would have been equivalent to around rents, but other factors are also at play. The introduction one-fifth of total per capita adult income, before factoring of free primary education in 2003 has not brought public in costs of uniforms, textbooks and informal fees. Given schooling to many informal settlements, leaving some of the large share of household budgets for the poor ab- the poorest households in the country with no alternative sorbed by food costs, it appears unlikely that the intake but to turn to low-fee—and low-quality—private provid- for median fee private schools was drawn from the poor- ers. Household surveys in informal settlements reveal a est half of Kenyan society, again calling into question the large unmet demand for public provision, with many poor merits of simple ‘public-versus-private’ comparisons. households turning to public providers when they are available. • Failure to examine underlying sources of cost-differences and implications for learning. The research exercise treats • Low-fee private schools are likely to remain an important cost-differences as a simple indicator of cost efficiency. part of the education landscape in Kenya. The growth of Detailed school survey evidence points to the need for these schools is in large measure a symptom of the failure greater nuance. Government owned schools register of public schools to provide the option of decent quality higher costs in part because they tend to have more text- education. However, it is not in itself evidence that the books per pupil, better buildings, high standards of water low-fee private sector is equipped to expand provision and sanitation, and more qualified teachers than low-fee of quality education on a more cost-effective basis than private providers. Driving down standards in these areas the public education system. This is especially true of the would hardly appear to be a desirable reform option in ASAL areas, where the market in private school provision the context strategies aimed at raising learning achieve- remains limited (see main text). For the vast majority of ment levels. The same is true for the primary source of Kenya’s children, especially the very poorest among them, the public-private school cost differential: namely teacher prospects for a decent quality education will continue salaries. While there are many problems with the train- to hinge on reforms that strengthen the equity and ef - ing, support and deployment of Kenya’s teachers, as ficiency of what is on any measure an under-performing well as with teacher absenteeism, driving down pay and public school system. conditions while seeking to increase and improve the quality of new career entrants in the name of efficiency is Sources: Bold et al 2011a; Bold et al 2011b; Glennerster et al 2011; Oketch and Ngware 2010; Ngware, Oketch and Ezeh 2011; likely to prove counterproductive. Oketch et al 2010.

with another national survey which found that half tended to perform better on numeracy (less so on liter- of Kenya’s teachers reported book-to-pupil ratios in acy). National average pupil-teacher ratios are margin- excess of 1-to-3 for English and math (Wasanga, Ogle ally above the guideline level of 40-1, but overcrowding and Wambua 2010). School infrastructure is another is a major problem in some areas. Moreover, the real concern that may impede quality of learning: Uwezo ratios may be far higher than reported because of found that four in 10 schools had no clean drinking teacher absenteeism. In 2011 the Uwezo survey found water and one in 10 no usable toilet. that 13 percent of teachers were absent from school at a time when they should have been present. Classroom overcrowding is another major concern. An econometric regression carried out in the 2010 Having teachers in the classroom is not an automatic National Assessment Center survey on learner guarantee of effective learning. The quality of class- achievement found that pupils in smaller classes room instruction experienced by many of Kenya’s

38 Global Economy and Development Program children leaves a great deal to be desired. Both the pupils registering low levels of learning achievement. National Assessment Center survey and the findings However, the ASAL counties account for a dispropor- of the Uwezo study point to weakness in teaching for tionately large share of Kenya’s national education basic literacy and numeracy as a national problem. deficit. Children from these counties carry disadvan- This raises questions about the quality and relevance tages associated with their home environment, includ- of teacher training. Teachers are poorly equipped to ing high levels of poverty, parental illiteracy and acute provide effective remedial teaching, even though it health problems. These disadvantages, especially occupies a significant share of their classroom time. female children, start before school and affect them In-service support does little to counteract the prob- throughout the education system, reinforcing wider lem. Around one-third of teachers reported no in-ser- cross county disparities. vice training between 2003 and 2009. Parental illiteracy has a marked bearing on prospects for school enrollment and learning. Children with Education Disadvantage in the 12 more educated mothers in particular are more likely ASAL Focus Counties to be in school and less likely to drop out. Having a The access and learning problems discussed in the literate home environment confers additional advan- previous section are evident across Kenya. Most of tages in terms of school preparedness and support the new counties have significant out-of-school popu- with homework. Most children in the 12 ASAL coun- lations and all have large numbers of schools and ties do not come from such an environment. These

Figure 10: Female Literacy and Gender Disparity (47 counties)


90 Kajiado Narok Lamu 80 West Pokot

70 Tana River Isiolo Marsabit 60

50 Samburu Garissa 40

Female/Male Ratio (percent) Female/Male 30 Turkana Wajir 20 Mandera


0 0 10 20 30 40 50 60 70 80 90 100 Female Literacy Rate (percent)

Source: KIHBS 2005. *Female literacy as a share of male literacy (Age 15+).

financing for a fairer , more prosperous kenya 39 counties occupy eight of the 10 bottom places in the parents and children in these counties in negotiating national ranking for literacy levels across Kenya’s 47 progression through the education system. counties. The gender disadvantage in adult literacy is particularly marked. In Samburu and Garissa fewer than half of females are literate, falling to less than Out-of-school Children one-third in Turkana, Wajir and Marsabit (Figure 10). As indicated by the enrollment data, the 12 ASAL counties account for a disproportionate share of ex- treme disadvantage in access to education. These The School Enrollment Deficit counties are home to 20 percent of the national pri- County-level data on enrollment highlights the gulf mary school-age population, but around 46 percent separating children in most of the 12 countries from of the out-of-school population. Put differently, being their peers across Kenya. The ASAL counties account born in one of the ASAL counties roughly doubles the for nine of the 10 lowest enrollment rates in the coun- risk of being out of school. try – and all 12 are in the bottom 15 (Figure 11). Turkana has the lowest net enrollment rate of any county, Stark as it is, even this figure understates the elevated with just one-quarter of primary school-age children risks facing some counties. Figure 13 compares the enrolled. That figure rises to just over one-third for share of each of Kenya’s 47 counties in the national Garissa and Wajir. Household poverty is likely to be a primary school-age population with the county share significant contributory factor in explaining the low of out-of-school children. In the case of counties such net enrollment rates in most ASAL counties. However, as Turkana, Wajir, Garissa and West Pokot the county the relationship between poverty incidence and share in the out-of-school population is more than school participation is nonlinear. Narok and Kajiado three times the population share. combine among the lowest poverty rates in Kenya with the lowest net enrollment rates. The data on out-of-school children draw attention to a wider set of challenges. Along with other countries, Part of the explanation for low overall enrollment can Kenya has adopted the Millennium Development Goals be traced to gender inequalities. The 12 ASAL counties target of universal primary education by 2015. The have some of Kenya’s deepest disparities in enroll- eliminating education charges in 2003 accelerated ment between girls and boys. Using the boy-girl ratio progress towards that target. However, the national to rank in primary school, the 12 ASAL counties are enrollment picture points to a marked slowdown included in the 13 counties with the largest gender gap since 2007. With almost half of Kenya’s out-of-school (Figure 11). Only West Pokot has a lower level of gen- children now concentrated in the 12 ASAL counties, der disparity below the national average changing this picture and getting on track for the 2015 target will require focused policy interventions Enrollment rates for secondary education in the 12 targeting these counties. This is an area in which the ASAL counties mirror those for primary school, with a constitutional commitment to affirmative action for magnified gender gap (Figure 12). Nine of the bottom the most marginalized counties and associated public 10 counties in the national ranking are in the group spending commitments could make a significant dif- of 12 ASAL counties for secondary school enroll- ference—an issue that we return to in Section 4 of ment. These figures illustrate the difficulties faced by this paper.

40 Global Economy and Development Program Figure 11: Primary School Ranking: Primary School Net Enrollment Rates and Gender Parity Ratios (47 counties, 2009)

Primary Net Enrollment Ratio 100 90 80 70 60 50 40 30 20 10 0 Kisii Kilifi Kitui Nyeri Meru Siaya Busia Embu Nandi Vihiga Kwale Migori Bomet Nairobi Nakuru Kisumu Kericho Kiambu Baringo Laikipia Nyamira Makueni Kirinyaga Murang'a Bungoma Mombasa Wajir (45) Wajir Homa Bay Isiolo (39) Machakos Lamu (33) Kakamega Nyandarua Narok (35) Trans Nzoia Trans Taita Taveta Taita Uasin Gishu Kajiado (34) Garissa (46) Turkana (47) Turkana Tharaka Nithi Marsabit (42) Mandera (43) Samburu (44) Tana River (40) Tana West Pokot (41) Pokot West National Average Elgeyo_Marakwet

Ratio of Primary Age Girls to Boys in Primary School 1.20






0.00 Kisii Kilifi Kitui Nyeri Meru Siaya Busia Embu Nandi Kwale Vihiga Migori Bomet Nairobi Nakuru Kisumu Kericho Kiambu Baringo Laikipia Nyamira Makueni Kirinyaga Murang'a Bungoma Mombasa Wajir (47) Wajir Homa Bay Isiolo (40) Machakos Lamu (37) Kakamega Nyandarua Narok (39) Trans Nzoia Trans Taita Taveta Taita Uasin Gishu Kajiado (36) Garissa (46) Turkana (43) Turkana Tharaka Nithi Marsabit (41) Mandera (45) Samburu (44) Tana River (42) Tana West Pokot (23) Pokot West National Average Elgeyo_Marakwet

Source: EMIS/Census 2009.

financing for a fairer , more prosperous kenya 41 Figure 12: Secondary School Ranking: Secondary School Gross Enrollment and Gender Parity (47 counties 2009)

100 Secondary Gross Enrollment Ratio 90 80 70 60 50 40 30 20 10 0 Kisii Kilifi Kitui Nyeri Meru Siaya Busia Embu Nandi Vihiga Kwale Migori Bomet Nairobi Nakuru Kisumu Kericho Kiambu Baringo Laikipia Nyamira Makueni Kirinyaga Murang'a Bungoma Mombasa Wajir (44) Wajir Homa Bay Isiolo (36) Machakos Lamu (35) Kakamega Nyandarua Narok (38) Trans Nzoia Trans Taita Taveta Taita Uasin Gishu Kajiado (17) Garissa (42) Turkana (47) Turkana Tharaka Nithi Marsabit (41) Mandera (40) Samburu (46) Tana River (45) Tana West Pokot (43) Pokot West National Average Elgeyo_Marakwet

1.2 Gender Parity (ratio of girls to boys in secondary school) 1.0





0.0 Kisii Kilifi Kitui Nyeri Meru Siaya Busia Embu Nandi Vihiga Kwale Migori Bomet Nairobi Nakuru Kisumu Kericho Kiambu Baringo Laikipia Nyamira Makueni Kirinyaga Murang'a Bungoma Mombasa Wajir (45) Wajir Homa Bay Isiolo (33) Machakos Lamu (34) Kakamega Nyandarua Narok (38) Trans Nzoia Trans Taita Taveta Taita Uasin Gishu Kajiado (12) Garissa (47) Turkana (44) Turkana Tharaka Nithi Marsabit (42) Mandera (46) Samburu (41) Tana River (43) Tana West Pokot (37) Pokot West National Average Elgeyo_Marakwet

Source: EMIS/Census 2009.

42 Global Economy and Development Program Figure 13: Kenya’s Unequal Distribution of Out-of-School Children (47 counties) 11

10 Mandera

9 Turkana 8

7 Wajir

6 Garissa 5

4 West Pokot Narok 3

2 Samburu Marsabit Kajiado

Share out of primary school aged children (percent) 1 Isiolo Lamu Tana River 0 0 1 2 3 4 5 Share of primary school age children (percent)

Source: EMIS/Census 2009.

Patterns of School Attrition The exercise reveals some contrasting patterns of Out-of-school numbers reflect the very high attrition school progression. Each of the four counties has a rates evident across the 12 counties. The odds are high level of dropout in the earlier grades, with con- firmly stacked against children making it through ba- tinued attrition across later grades. In each case, the sic education, with those who succeed facing another number of children sitting in Grade 8 classrooms is set of barriers at the point of transition to secondary less than one-half of the number in Grade 1: in Turkana education. and Wajir it is around one-quarter. Prospects for pro- gression through the point at which children sit the

As is the case at the national level, data constraints KCSE are highly unfavorable. While children across make it impossible to construct longitudinal cohort- the four counties share a limited likelihood of reaching tracking exercises for the 12 counties. Using 2010 data secondary school, dropout patterns vary. In Turkana, made available by the Ministry of Education, we con- the number of students in Standard 3 is less than half struct a proxy tracking exercise by mapping numbers the number entering Standard 1, whereas West Pokot enrolled by grade for four of the ASAL counties, start- registers a less steep decline. Progression profiles ing with the cohort that entered Standard 1 in 2003 also vary by gender. The disparities are limited in West (Figure 14). Pokot and Turkana, but far wider in Wajir and Garissa.

financing for a fairer , more prosperous kenya 43 Both the high overall level of early grade attrition and directed towards countering the pressures leading to the differences between counties have implications elevated risk of dropout during the primary cycle. for education financing. If closing county-level gaps in progression towards universal primary education The cumulative effect of attrition in primary and is a core policy goal, the financial support has to be secondary school can be best seen at the point that

Figure 14: School Progression Profiles: Enrollment Levels by Grade for Wajir, Turkana, West Pokot and Garissa (2009) Wajir 8000


6000 Total 5000 Female

4000 Male

Enrollment 3000



0 Std 1 Std 2 Std 3 Std 4 Std 5 Std 6 Std 7 Std 8 Form 1 Form 2 Form 3 Form 4 KCSE Grade

Source: EMIS 2010.

Turkana 25000


Total 15000 Female


Enrollment 10000


0 Std 1 Std 2 Std 3 Std 4 Std 5 Std 6 Std 7 Std 8 Form 1 Form 2 Form 3 Form 4 KCSE Grade

Source: EMIS 2010.

44 Global Economy and Development Program West Pokot 25000



15000 Female


Enrollment 10000


0 Std 1 Std 2 Std 3 Std 4 Std 5 Std 6 Std 7 Std 8 Form 1 Form 2 Form 3 Form 4 KCSE Grade

Source: EMIS 2010.

Garissa 16000


12000 Total 10000 Female 8000 Male

Enrollment 6000



0 Std 1 Std 2 Std 3 Std 4 Std 5 Std 6 Std 7 Std 8 Form 1 Form 2 Form 3 Form 4 KCSE Grade

Source: EMIS 2010.

Kenya’s children sit for the KSCE. Figure 15 provides effort). In other words, the two sides in Figure 15 would a simple comparison of the share of each of the new be of equivalent length. With 18.5 percent of the sec- counties in the secondary school-age population and ondary school-age population, the 12 ASAL counties their share of pupils sitting the KCSE. This is a very would account for a similar proportion of KCSE exam rough measure of equity, but it is nonetheless telling. candidates. They account for 5 percent of KCSE candi- In a situation of equal opportunity, the distribution of dates. Gender disparities are very large across the 12 students sitting the KCSE would mirror the distribu- ASAL counties, with girls representing around 4.2 per- tion of the eligible population (adjusted for chance and cent of female candidates that sat for the 2010 exam.

financing for a fairer , more prosperous kenya 45 Figure 15: Unequal Opportunity: Share of County Secondary School-Age Population and Share of Cohort Sitting the KCSE (2010)

Isiolo Tana River Samburu Mandera accounts for 4 percent Lamu of the secondary school age children, Marsabit but just 0.4 percent of the KSCE candidates. (Girls account for 0.1 percent of the KSCE Tharada Nithi candidates) Wajir Turkana Mandera Garissa West Pokot Taita Taveta Narok Kwale Elgeyo Marakwet Kajiado Laikipia Baringo Busia Kilifi Share of male students Mombasa sitting for KCSE Kericho Trans Nzoia Share of female students Embu sitting for KCSE Nandi Uasin Gishu Share of the national secondary Kirinyaga school age children Nyamira Nyandarua Migori Vihiga Siaya Kitui Homa Bay Bomet Kisumu Nyeri Machakos accounts for 3 Makueni percent of the secondary Bungoma school age children, but just Kakamega 4 percent of the KSCE candidates. Machakos (Girls account for 2 percent of the Murang'A KSCE candidates) Meru Nairobi Nakuru Kisil Kiambu

10 5 0 5 10 KCSE Students Secondary school age population

Source: KNEC: KCPE 2010.

46 Global Economy and Development Program Learning Achievement KCPE candidates in 2010, which is under half of their Getting through to the exam stage of the education collective share in the primary school-age population. cycle is an indicator of school progression, not of learning achievement. How do pupils from the ASAL Bearing in mind the sample size caveat, the 12 ASAL counties perform in the KCPE and KCSE exams rela- counties have a mixed record on exam performance. tive to their peers from other counties? On a simple ranking of mean scores for the KCPE, eight of the 12 ASAL counties are in the bottom 20,

That question has to be addressed with some cau- with Mandera, Tana River and Garissa in the bottom tion. Given that the vast majority of children entering five (Table 5). Mandera registers the lowest test score school drop out long before reaching the relevant of any county. At the other end of the spectrum, two KCPE let alone the KCSE grades, the exams provides counties – Kajiado and West Pokot – are in the top a reference point for a very small sample of children. quartile of counties, with Turkana and Samburu in the The vast majority have dropped out before taking top half of the distribution. the exams, presumably at far lower levels of learn- ing achievement. As a group, the 12 ASAL counties Figure 16 looks beyond the county average perfor - covered in this report accounted for just 8 percent of mance to the test score distribution for four counties.

Table 5: Kenya Certificate Primary Examination Average Scores: 12 ASAL Counties and National Average (2010) Rank (out of Gender County Mean 47 counties) Male Female disparity Mandera 218 47 223 207 0.96 Garissa 220 44 222 214 0.93 Tana River 220 46 227 208 0.98 Lamu 226 42 229 222 0.97 Wajir 232 38 233 228 0.93 Marsabit 240 34 250 224 0.90 Narok 242 32 249 232 0.93 Isiolo 242 31 250 232 0.90 Samburu 253 17 262 237 0.92 Turkana 254 15 259 245 0.95 Kajiado 258 11 261 254 0.98 West Pokot 267 6 272 261 0.96 ASAL Average (12 counties) 239 - 245 230 0.94 National 247 - 252 240 0.95

financing for a fairer , more prosperous kenya 47 Figure 16: Distribution of KCPE Test Scores: Turkana, West Pokot, Tana River and Mandera (2010)

County: : Tana River Poor Pass Very Good Poor Pass Very Good 3000 2364 2500 2431 2440 2187 2000 2000 1455

1500 1334 Frequency Frequency Frequency Frequency 1000

1000 749 631 520

500 280 129 26 54 86 0 0 0 50 100 150 200 250 300 350 400 450 500 0 50 100 150 200 250 300 350 400 450 500 Score Score

County: : Turkana Poor Pass Very Good Poor Pass Very Good 2000 2364 1014 1799 1000 880 1580 1500

680 625 1000 Frequency Frequency Frequency Frequency 500 708

234 500 234 195 216 24 54 24 0 0 0 50 100 150 200 250 300 350 400 450 500 0 50 100 150 200 250 300 350 400 450 500 Score Score

Source: KNEC: KCPE 2010.

It illustrates the diversity of performance. While very Counties such as Turkana perform well on test scores, few children in West Pokot and Turkana take the KCPE, in part because such as small proportion of students those that do perform relatively strongly with just make it through to the KCPE. For counties in this 7 percent and 15 percent respectively scoring below category the priority is to maintain learning achieve- 200 (well below the national average of 24 percent). ment levels while increasing the number of children This is in marked contrast with the situation in Tana progressing to Grade 8. In the case of counties like River and Mandera, where the proportions scoring Mandera, the desperately low levels of learning below 200 are respectively 41 percent and 39 percent. achievement among KCPE points to fundamental fail- ures in the education system, allied to wider pressures These figures draw attention to the twin challenge that lead children to drop out. in access and learning facing the ASAL counties.

48 Global Economy and Development Program There is a consistent pattern of boys outperforming for boys and girls in the 12 ASAL counties are just un- girls in KCPE scores across the 12 ASAL counties, der 3 percent and 1 percent. Only boys in West Pokot although the gender gap varies across counties. For perform above the national average. Similarly, the instance, Wajir has very high levels of gender disparity chances of children from the ASAL counties scoring in school participation, but a lower level of disparity in above a C grade are well below the national average. test score; Samburu and Isiolo have levels of dispar- ity in test scores well above the national average. The Gender disparities in KCSE scores are far wider across persistence of the gender disparities in exam scores the 12 ASAL counties than the rest of Kenya. Girls are even for girls who make it through to the KCPE points less than half as likely as boys to score a B+. While to serious concerns that reflect the learning environ- the gender gap is narrower for C+ performance it is ment they face at home and at school. still wide: the gender parity ratio is 0.68. Here, too, there are some marked variations. Girls in Turkana Results at the KCSE level are more discouraging for and Isiolo have far less chance than boys of making the ASAL counties (Table 6). Students seeking to se- it through to the KCSE, but those that do perform al- cure funding for progression into higher education most on par with boys in achieving a score of B +. By in Kenya are required to achieve a B+ score. In 2010, contrast, girls in Garissa are half as likely to score at 7 percent of boys and 4 percent of girls across the B+, falling to less than one-fifth as likely in West Pokot. country achieved that grade. The comparable figures

Table 6: KCSE Results: Selected ASAL Counties and National Average (2010) KCSE 2010 Male (%) Female (%) Gender Disparity County Share (%) B+ C+ B+ C+ B+ C+ School-Age County Above Above Above Above Above Above Population Candidates Garissa 1.0 19.1 0.5 8.6 0.47 0.45 1.8 0.5 Isiolo 1.1 15.6 1.1 13.0 0.96 0.83 0.3 0.2 Kajiado 4.7 25.9 3.5 27.8 0.74 1.07 1.5 1.2 Lamu 2.5 17.0 0.3 16.4 0.12 0.96 0.2 0.2 Mandera 1.9 16.0 0.0 3.6 0.00 0.22 3.6 0.4 Marsabit 1.9 23.0 1.5 11.2 0.79 0.49 0.8 0.2 Narok 1.4 19.3 0.9 15.6 0.61 0.81 2.0 0.9 Samburu 3.0 34.8 1.5 23.2 0.50 0.67 0.6 0.2 Tana River 0.5 5.9 0.0 2.9 0.00 0.49 0.5 0.2 Turkana 3.2 28.5 3.1 26.6 0.94 0.93 2.9 0.4 Wajir 0.2 19.4 0.0 2.8 0.00 0.14 2.2 0.3 West Pokot 9.9 46.0 1.9 33.8 0.19 0.73 1.4 0.6 ASAL Average (12 counties) 2.6 22.6 1.1 15.46 0.45 0.68 1.54 0.44 National 7.1 33.1 4.4 26.9 0.62 0.81 - - Source: KNEC: KSCE 2010.

financing for a fairer , more prosperous kenya 49 Uwezo Survey Evidence Disparities within the 12 counties are significant in Looking beyond exam results the absolute level of some cases. Within , the proportion learning in the ASAL counties is very low. The Uwezo of children in Standard 3 able to read a Standard 2 survey results for the counties suggest that large Kiswahili text ranges from 14 percent in to numbers of children progress through the primary 73 percent in Kajiado North (the area close to Nairobi). school system without acquiring basic competencies At the other end of the average performance scale, —a state of affairs that inevitably contributes to high in the share of children in Standard levels of dropout. 3 able to do a Standard 2 division ranges from 12 per- cent in Turkana Central to 4 percent in Turkana South.

Taken as a group, the ASAL counties perform far be- Once again, these disparities underline the impor - low the national average performance levels on basic tance of subcounty-level data in identifying areas of competencies (Figure 17). The gap is evident as early acute deprivation. as Grade 3. The Uwezo national ranking covers 124 districts by the proportion of Grade 3 children able to Private Schools: A Limited Presence achieve Grade 2 standards for literacy and numeracy. The 12 ASAL counties account for 20 of the bottom 24 Private schools have a limited presence in most of the districts. In Turkana county just 8 percent of children 12 ASAL counties. In most cases underlying conditions in Standard 3 could perform a Standard 2 division. are not favorable to the development of a market for Only one in 10 children in Standard 3 in Wajir could private providers. Low average-incomes, high levels of perform at Standard 2 levels. poverty and low population density all limit demand, while the small number of students reaching the early grades of secondary school limits the supply of poten- Figure 17: Unequal Achievement: Learning tial teachers. Levels for Arid Districts and Average for All Districts (2011) The Uwezo survey provides a useful point of com- parison. Nationally, it reports that around 12 percent 70 of children were covered by private providers in 60 2011, with urban areas registering the highest con- 50 centration. By contrast, just 1-2 percent of children in Samburu, Turkana, Wajir and Tana River were re- 40 ported as attending private schools, rising to 3-5 per- 30 cent in Mandera and Isiolo. Only Kajiado exceeded the 20 national average, with 20 percent of children enrolled Number of children (percent) in private schools. 10

0 Class 3 children who Class 3 children who Children in the 12 counties were also far less likely to can read a paragraph can do class 2 subtraction be receiving private tuition. While 67 percent of par- ASAL Districts National ents of children covered by the Uwezo study in Nairobi Source: Uwezo 2011. reported receiving private tuition, the comparable

50 Global Economy and Development Program shares for Turkana, Samburu and Wajir ranged from activities can compete with schooling. While hard data 6-12 percent. is lacking, it is likely that drought has damaging ef- fects on education in the ASAL counties. It drives up Several policy conclusions can be drawn from the pat- the price of food, contributing to child malnutrition tern of private provision and household expenditure and ill-health, leads to women and girls spending more on tuition. The limited presence of private schools time collecting water, and results in young boys herd- implies that public schools will have to play the cen- ing over longer distances. tral role in addressing problems of access and quality. Leaving aside the wider debate over the quality and Even without drought distance is a major concern. cost effectiveness of private schools, it would appear In counties with low population densities boarding unlikely that most of the ASAL counties will develop schools may offer the only prospect of a secondary a private school market of any scale in the near-term education, or even the upper years of basic education. future. Given the limited household expenditure on However, the costs of attending boarding schools are private tuition, which is in part a corollary of high lev- prohibitive for the vast majority of people living in the els of poverty, there are also strong grounds for pro- 12 counties. Pastoralist livelihoods can also present a viding public spending increments to schools in ASAL challenge. While education planners tend to think of targeted at raising learning standards. schools as a fixed structure, pastoralist herders travel over long-distances. Given that pastoralist children start herding in many cases before their adolescent Barriers to Enhanced Access and years, the implication is that either school terms have Quality to adjust, or schools themselves have to become mo- All of the barriers to access, retention and learning bile (Krätli and Dyer 2009; Government of Kenya 2010). identified for Kenya appear in concentrated form in the 12 ASAL counties. The effects of household pov- Learning prospects are further impaired by school- erty, food insecurity and parental illiteracy weigh based factors. In counties such as Wajir, Garissa and heavily on children’s education prospects long before Mandera, over 40 percent of children attend schools they enter school. lacking desks and chairs, with children sitting on the floor (Uwezo Kenya 2011). Pupil-teacher ratios are Pupil absenteeism rates provide a barometer of wider very high in some areas. Both Turkana and Mandera disadvantages affecting education. At the time of the have ratios above 50-1. In another five counties — 2010 Uwezo survey over 40 percent of enrolled chil- Wajir, Tana River, Marsarbit, Narok and West Pokot— dren were reported out of school in Narok, Samburu, the ratio is between 40 and 50-to-1. Given that many Tana River, Turkana and Wajir (Uwezo Kenya 2011). of the children entering schools in the ASAL counties Ill-health is a major contributory factor, with malaria are first generation learners from non-literate home and nutrition-related conditions the most prevalent environments requiring special support, these are problems. Livelihood factors also weigh heavily. Young very high ratios. Factoring in teacher absenteeism boys from pastoralist homes take on early responsibil- would inflate the ratio in many counties. In Turkana, ities for herding, while young girls are intensively en- one in five teachers was absent on the day of the gaged in the collection of water and firewood. These Uwezo survey team visit in 2010 (Uwezo Kenya 2011).

financing for a fairer , more prosperous kenya 51 The ASAL counties also face wider difficulties. As in background make recruitment, deployment and re- other counties, the quality of the teacher workforce tention of teachers difficult” (Government of Kenya is compromised by a training system geared towards 2010). While the Council has to yet be established it is rote learning and by inadequate in-service support. provided for in the Education Bill currently before the Beyond these general problems, concerns have Kenyan Parliament. been raised over the relevance of the national cur- riculum for children in the ASAL counties, especially Current policy approaches appear insufficient to those from pastoralist households (Commonwealth address these problems, some of which are self- Secretariat 2007). There is also evidence that schools reinforcing. To take one example, the limited flow of in counties characterized by low population density pastoralist girls into secondary education limits the struggle to recruit and retain experienced teachers. supply of potential teachers. The policy implication is This has been recognized in a number of national that the difficulties associated with teacher recruit- education strategy documents. In 2008 the govern- ment and retention has to be addressed partly in the ment of Kenya made a commitment to establish a education system through a strengthened focus on National Council for Nomadic Education, prompted in the retention of female students, and partly through part by a recognition that “the hardships associated the teacher management system. with the ASALs and the few teachers with a nomadic

52 Global Economy and Development Program Some Lessons from assignment and intergovernmental transfers. International Experience Countries vary enormously in the degree to which they devolve responsibilities for expenditure and rev- he marked inequalities across Kenya’s counties enue mobilization, though almost every country in Toutlined in previous sections are the results of the world makes some attempt to mitigate horizontal many factors. Historical legacy, patterns of economic disparities between different parts of the country growth and political marginalization have all contrib- through intergovernmental transfers (Bahl 2010). uted. Public spending patterns have also played a role, with more commercial farming areas and urban cen- Horizontal disparities can arise for many reasons. In ters capturing the lion’s share of budget allocations. countries that have highly devolved revenue systems, One of the aims of the public spending provisions in fiscal imbalances can arise because of differences in the new constitution is to counteract the horizontal average income (and hence the tax base) of richer inequalities between counties and the vertical im- and poorer areas (Bird and Bahl 2008). On the ex - balances between groups that characterize Kenyan penditure side of the equation, the costs of delivering society. basic services can also vary as a result of differences in terrain, population density or distance from high- Kenya is not the only country addressing this issue. ways. When subnational entities are responsible for Governments across the world use intergovernmental delivering basic services, the financing requirements transfers, targeted support measures and national for achieving a national minimum standard of provi- programs aimed at mitigating national inequalities, sion will also reflect inequalities in access. Regions and at establishing a minimum level of basic service that are furthest from the required level of coverage provision. Some of these programs transfer revenues will face higher financing requirements. Formulae for from central to devolved governments. Others target determining intergovernmental transfers will typically specific forms of deprivation by targeting identifiable include (i) norms for the provision of basic services groups, regions or individuals. The design of programs (ii) an assessment of the fiscal capacity of the govern- and broad national approaches reflect the different ment entity charged with providing the service (iii) institutional arrangements, political processes and criteria for establishing current shortfalls in provision history of different countries. There are no blueprints and (iv) the estimated costs of achieving specified for Kenya to draw on. By the same token, international goals (Bahl 2008; Bahl and Wallace 2004). experience offers some useful guidelines and lessons that may have relevance for the debate over equitable Approaches to the correction of horizontal and verti- sharing of public finance in Kenya. cal imbalances in Kenya will be shaped in part by the country’s distinctive model of devolution. That model Intergovernmental Transfers is marked by elements of continuity as well as change. One such element is the degree of centralization in Fiscal and political decentralization is fundamentally the fiscal system with respect to revenue mobiliza - an exercise in transferring budget authority and tion. Some 24 percent of Kenya’s GDP is mobilized as devolving decision-making to subnational levels of central government revenue, with local government government. Broadly, the process revolves around sources—mainly property taxes—accounting for 1-2 three practices: expenditure assignment, revenue

financing for a fairer , more prosperous kenya 53 percent of GDP (World Bank 2011b). Nairobi City alone administrative strains. In the midst of these uncertain- accounts for just over 40 percent of own-source rev- ties, the Commission for Revenue Allocation (CRA) enue for municipalities—giving the city a per capita has been charged with developing formulae for inter- revenue base some 20 times higher than those re- governmental transfers and wider budget allocations ported in poorer rural municipalities, such as those in that will enact the principles of the new constitution. the ASAL counties (Government of Kenya 2010). The South African experience has a special relevance Devolved financing will increase the level of intergov- for Kenya. Parts of the 2010 constitution draw heav- ernmental transfers. The constitution mandates that ily on South Africa’s constitutional arrangements. a minimum of 15 percent of national revenue is to be Moreover, South Africa has one of sub-Saharan transferred on an unconditional basis to the counties Africa’s most highly developed systems of intergov- to cover their responsibilities assigned to them. This ernmental transfers aimed at reducing horizontal is an increase over the financing provided through inequalities. Decentralized financing has been one the currently devolved funds, the Constituency element in a wider set of fiscal measures aimed at Development Fund and the Local Authorities Transfer combating the legacy of apartheid. Fund, which accounted for around 3 percent of the an- nual budget in 2009 (World Bank 2011b). Importantly, Central government allocations to devolved authori- the constitutional provisions governing devolved fi- ties in South Africa are determined by a ‘Provincial nancing extend not only to devolved financing but to Equitable Share’ formula that attaches varying the overall public spending envelope. This implies that weights to population and equity goals. For example, future governments could be required to demonstrate the transfers for health are determined by population an intention to correct horizontal and vertical imbal- size and by the size of the population without access ances through the 85 percent of the budget falling to medical aid. In education, the size of the school-age outside of the devolved funds. population is adjusted in the funding formula by the size of the out-of-school population. As we suggest in the next section of this paper, this is an approach that International Experience would help to develop more equitable financing in ed- Any assessment of international experience and its rel- ucation for Kenya. The same is true of the provisions evance for Kenya has to take into account the specific made in South Africa’s intergovernmental transfer characteristics of the country’s path to devolution. system for the weighting of poverty (Box 3). Unlike countries such as India or Brazil, Kenya will not have subnational entities with strong revenue raising More devolved fiscal systems also hold out lessons. powers and high levels of fiscal autonomy. Moreover, The case of India is instructive because of the coun- given the highly centralized nature of both the fis - try’s long experience in the design, development and cal system and the political system, Kenya does not implementation of intergovernmental transfer sys- have well-defined norms, rules and institutions for the tems. Here too, equity has been a central theme in governance of subnational entities. The creation of determining transfers to states. One of the features of 47 new counties out of the old system of 158 districts Indian federalism is the use of a formula to determine assembled in eight provinces will create political and the fiscal capacity of states, taking into account their

54 Global Economy and Development Program Box 3: South Africa’s Provincial Equitable Share • A basic share (14 percent) derived from each province’s (PES) Approach share of the national population.

More equitable public spending was identified as a prior- • An institutional component (5 percent) divided equally ity by the post-apartheid government in South Africa. The between provinces. country’s experience is of direct relevance to Kenya be- cause both central and local governments operate under • A poverty component (3 percent) based on the percent- a constitution enshrining a strong commitment to equity age of people residing in the province living below the in service provision. poverty line.

• An economic component (1 percent) based on GDP by Over the past 15 years South Africa has developed a region. complex system of intergovernmental transfers aimed at promoting greater equity across regions and social Some elements of the formula are overtly redistributive. groups. That system operates through formulae that at- Provinces such as Eastern Cape, Limpopo and KwaZulu- tach considerable weight to identified sources of social Natal receive larger shares of the poverty, health and disadvantage. education budgets than their basic share, while more pros- perous provinces with better indicators receive less. The Provincial Equitable Share (PES) budget is at the heart of the devolved financing system. Allocated by the The PES system has been subjected to periodic review central government, this accounts for over 80 percent of and extensive critical scrutiny. Various weaknesses have provincial government revenue. The PES transfer oper- been highlighted. For instance, financing provisions are ates through a formula that is updated annually. For the not linked to detailed estimates of the costs of delivering 2008 budget, the distribution of weights by component basic services in particular provinces, raising concerns was as follows (Alm and Martinez-Vasquez 2009): over equivalence in provision. In the case of education, the higher per capita costs associated with reaching and deliv- • An education share (51 percent) based on (i) the size of ering effective learning to highly marginalized populations, the school-age population and (ii) the number of learn- coupled with the presence of over-age children repeating ers enrolled in public schools. Each component is as- secondary school grades, may disadvantage the poorer signed a weight of 50 percent. provinces. Lastly, the weighting for household poverty is seen by some as too low. • A health share (26 percent) based on (i) overall popula- tion and (ii) the proportion of the population without ac- Sources: Rao and Khumalo 2004; Petchey et al 2007; Financial and Fiscal Commission 2009. cess to medical aid. The weighting for (ii) is four-times that for (i).

highly unequal average income levels (Box 4). An ex- limited revenue raising powers, this is a principle that plicit goal is to narrow the gap in fiscal capacity, and has direct relevance for the country – not least in the hence capacity to deliver basic services, between the light of the very unequal levels of service provision poorest and richest states. An underlying principle of now in evidence. the Indian intergovernmental transfer system is that all states should be in a position to provide compa- Not all equalization measures operate through the rable levels of public services despite differences in general system of intergovernmental transfers. their revenue raising capabilities (Chakraborty 2010a; Provisions for specific sectors can also seek to redress Chakraborty 2010b). While Kenya’s counties will have horizontal imbalances, with specific programs that

financing for a fairer , more prosperous kenya 55 Box 4: Intergovernmental Transfers in India allocations. There is an inverse correlation between aver- age state-level income and per capita transfers from the India seems an unlikely point of reference for comparison national tax revenue pool to state governments. For exam- with Kenya on the issue of financial devolution. With the ple, the poorest state (Bihar) receives almost three times world’s largest population, a highly devolved political sys- as much on a per capita basis as the richest (Haryana). tem and decentralized financing arrangements that have While the transfer does not equalize revenues, it mitigates evolved over more than six decades, the country has a long the revenue gaps and the resulting differences in financing track record in developing arrangements for intergovern- capacity across India’s states. Notwithstanding the marked mental transfers. Even so, an awareness of India’s arrange- differences in national contexts, this is an arrangement ments could help to inform public debate in Kenya. that merits some consideration in Kenya. Wider aspects of the debate surrounding the intergovernmental transfer One of the unique features of devolved financing in India system in India may also be relevant for Kenya. Critics point is the role of the Union Finance Commission (UFC). This out that area and population are neutral indicators of need, is a constitutional body charged with correcting vertical and that the fiscal discipline provisions can have the effect and horizontal imbalances in financing. Recommendations of limiting expenditures without reference to need. While of the UFC have a near-binding status with respect to the the horizontal transfer formula mitigates fiscal disparities system of intergovernmental transfers. These transfers are between states, it does not eliminate the very large in - significant. They represent around 5 percent of GDP, or just equalities in expenditures on basic services and economic under half of central government revenue. India’s states infrastructure associated with wealth disparities. raise around 8 percent of GDP through their own revenue mobilization efforts. The system of intergovernmental transfers is just one component in a wider set of transfers. Central government Rapid economic growth in India has been associated with also finances a wide range of sector and state-specific pro- persistent and widening inequalities, including interstate grams. These include grants for the flagship national edu- disparities. The horizontal distribution formula for 2010- cation program – the Sarva Shiksha Abhiyan (SSA) – that 2015 was intended to redress inequalities in fiscal capac- provides states with the capacity to meet their obligations ity between middle-income and high-income states (such under the 2009 Right to Education Act, transfers for the as Maharashtra, Haryana and Gujarat) and poorer states Mahatma Gandhi Rural Employment Guarantee Program (such as Bihar, Chattisgarh and Uttar Pradesh). The for- and national programs on child nutrition. One of the criti- mula incorporates four indicators, each with a different cisms leveled against the wider public financing architec- weight attached to it: fiscal capacity (47.5 percent), popu- ture is that there is no integrated structure capturing the lation (25 percent), fiscal discipline (17.5 percent) and area degree to which overall resources are allocated against (10 percent). needs.

The fiscal capacity provision is aimed at increasing the There are at least two features of the Indian model that resources available to low-income states with a limited tax have some resonance with debates in Kenya. The first base relative to higher-income states. It does so through a concerns weighting. Under the current horizontal financ- formula that uses the average tax-GDP ratio by state as a ing formula, India attaches a modest weight (25 percent) norm for determining the ‘potential tax revenue’ of each to population – and far less than in Kenya. Conversely, the state based on its income level. This is used to calculate the weight attached to fiscal capacity equalization reflects the ‘fiscal distance’ between this potential and the revenues concern in establishing comparable levels of public service that would be generated in the highest-income state ap- for comparable levels of taxation. This affirmative financ- plying the same tax rate. The gap, or the ‘fiscal capacity ing arrangement that favors poorer states reflects the distance’ as it is known, determines just under half of the spirit of Kenya’s 2010 constitution. tax transfer from central government.

Second, intergovernmental transfers are one part of a As illustrated in the figure below, the ‘fiscal capacity dis- wider set of financing instruments used in India. The rapid tance’ formula has a markedly equalizing effect on revenue

56 Global Economy and Development Program progress that India has registered in basic education re- and disadvantage holding back the arid and semi-arid coun- flects the combined effects of programs such as the SSA, ties in Kenya. rural employment guarantees and other measures that Sources: Chakraborty 2010a; Chakraborty 2010b; Isaac and target highly marginalized populations. Such programs Chakraborty 2008. could play a critical role in breaking the cycle of poverty

Per Capita Tax Devolution of General Category States: 2009-2010

2300 2100 Orissa Jharkhand 1900 Bihar Chhattisgarh 1700 Madhya Pradesh 1500 Uttar Pradesh Andhra Pradesh 1300 Rajasthan West Bengal Tamil Nadu Karnataka Kerala 1100 Gujarat 900 Haryana 700 Punjab Maharashtra 500 15000 25000 35000 45000 55000 65000 75000 85000 95000

Per Capita Income in Ascending Order

Sources: State Finances: A Study of Budgets of 2011-12, Reserve Bank of India;

target vertical income imbalances playing a supple- education financing through intergovernmental trans- mentary role. fers while increasing demand for education through cash transfers can greatly enhance access and learn- One example comes from the education sector in ing achievement levels (Box 5). This is a lesson that Brazil. Financing for the sector is highly devolved, with has some resonance for Kenya given that horizontal local taxes the dominant source of revenue. However, and vertical inequalities in education have stalled states in the poorer northeast have less revenue-mo- progress towards the 2015 MDG targets. bilization capacity than those in the wealthier south. The resulting disparities in per pupil financing ca- Most developed countries also make provisions for pacity are partially mitigated by central government redressing education disparities through national transfers from the federal education budget, with financing. One illustration comes from the United the Bolsa Familia social protection program providing States, where the federal governments Title 1 program cash transfers to poor households on the condition for primary and secondary schools allocates resources that they keep children in school. One of the lessons for measures that target poor performing schools from Brazil is that narrowing horizontal disparities in characterized by high levels of poverty. Schools that

financing for a fairer , more prosperous kenya 57 Box 5: Redistributive Public Finance in Brazil: intergovernmental transfers significantly reduce the fi- Equalizing Opportunity in Education nancing disparity. Education inequalities are a major source of poverty and Other programs also act to reduce intrastate disparities. wider social disparities in Brazil, with extreme wealth The Bolsa Familia program (see text) transfers 1-2 percent differences between states reinforcing socio-economic of Brazil’s GDP to around 11 million of the country’s poor- fault lines. Reforms over the past decade have sought to est households, many of which live in the northern states. mitigate horizontal disparities between states, while social protection programs have sought to break the link between Transfers are conditional on school attendance—and there poverty and educational disadvantage. is evidence that they have significantly increased demand for schooling.

While revenue mobilization is decentralized, education financ- ing in Brazil is managed through central government norms. More equitable financing, allied to wider institutional re- The federal government uses a national formula to stipulate form, has transformed the state of Brazilian education. the share of state taxes that have to be assigned to education. The wealth gap in school attendance has narrowed: chil- Government norms also stipulate minimum levels of spend- dren for the poorest quintile now average eight years in ing per pupil for each level of education, with higher levels of school compared to four years in the mid-1990s. Learning financing required for rural areas and disadvantaged groups achievement levels have also improved. The 2009 such as indigenous people and black Brazilians. Program for International Student Assessment recorded a 52 point increase in Brazil’s math score since 2000 – equivalent to gaining a full academic year and one of the Wealth disparities mean that states vary in their capacity fastest increases on record. to mobilize resources. Average income in poor northern states such as Para, Cerea and Maranhao is less than half of the level in richer southern states such as Rio Grande do While the differences have to be recognized, the margin- Sul and Sao Paulo. Without central government transfers alization and low levels of learning achievement in Kenya mirror those of Brazil a decade ago. The lesson from Brazil through an education financing facility—the Fundeb—sev- is that more equitable finance linked to policies aimed at eral poorer states would be unable to meet the required strengthening national learning assessment systems, tar- levels of spending. These transfers amount to around one- geting under-performing schools, pupils and regions, and fifth of state spending on education in Ceara, rising to more improved teacher training can deliver the type of results than one-third in Para and Maranhao. that Kenya aspires to in the Vision 2030 strategy

While these transfers do not equalize spending—per pupil Sources: Henriques 2009; UNESCO 2010; Bruns, Evans, and Luque 2012. financing in Sao Paulo is twice as high as in Maranhao—

enroll at least 40 percent of children from low-income While the circumstances are clearly very different to households are eligible, along with schools operating those prevailing in Kenya, these two cases illustrate a programs aimed at identifying and supporting ‘failing broader principle. In both cases the objective is to pro- children’ (Lefkowits 2004). Another example is the vide higher levels of per capita financing for pupils facing United Kingdom’s ‘relative needs’ formula, which is identifiable disadvantages. Put differently, the recogni- used to determine the central government grant to tion that equal spending for children in unequal circum- local government. The formula includes a ‘pupil pre- stances is not compatible with a commitment to equity. mium.’ Tied to the number of children eligible for free school meals, a proxy for household deprivation, this Several countries have adopted local government raises funding per pupil by 50 percent (Government of financing formulae with an explicitly redistributive the United Kingdom 2011). bias in favor of disadvantaged regions. Many of these

58 Global Economy and Development Program formulae also introduce a proxy weighting for the cost One of the distinctive features of the Ethiopian ar- of service provision. Examples include the following: rangement is that it establishes what is effectively a basic service minimum entitlement for all citizens, • Tanzania has a nonsectoral capital development with the intergovernmental transfer system geared grants program with allocations weighted by popu- lation (70 percent), territory size (10 percent), towards delivering that entitlement. Given that one of poverty count (20 percent) and local government the aims of Kenya’s Equalization Fund is to raise basic performance (20 percent). services in marginalized areas to the standards en- joyed by the rest of the country, a starting point might • In 2006 Rwanda adopted an allocation formula for be to adapt the Ethiopian approach by defining a mini- block grants to local government based on popula- tion (20 percent), a proxy for poverty (20 percent, mum standard and estimating the costs of meeting using revenue collection), size of area (10 percent) that standard on a county-by-county basis. and an estimated financing gap between revenue collection and costs of administration (40 percent). Several countries have used regional development

• Zambia provides recurrent financing grants through policies to target support towards regions—or sub- a formula that includes population weighted by a regions—and groups characterized by high levels of deprivation index giving equal weight to the follow- disadvantage. The large-scale regional development ing: number of poor people, the percentage of the strategies adopted by China (in western regions) and population lacking access to water, sanitation, mar- Brazil (for the northeast), programs in India target- kets and public transport. ing scheduled castes and tribes, and an Indonesia • Nepal operates a system of government-financed de- program for less-developed villages are all examples. velopment grants allocated on the following criteria: One model that may have some relevance for Kenya population (50 percent), the Human Development is Vietnam’s ‘Program 135,’ which has identified com- Index (10 percent), size of territory (10 percent) and munes in the most disadvantaged provinces for sup- a weighted cost-of-service index (25 percent). port in areas such as health, nutrition and education (Box 7). The experience of Ethiopia is of considerable inter- est in the Kenyan context, not least because of the country’s relatively recent transition to devolved Cautionary Tales financing. The intergovernmental transfer operates While almost all countries have embraced the prin- through a grant allocation formula developed by a ciple of horizontal equalization in their approach to technical committee and approved by parliament (Box intergovernmental transfers, many have struggled to 6). As in India, the formula includes a methodology translate principle in practice. Uganda is a case in point. for estimating the potential tax base of each region. Recognizing the degree to which some districts were Unlike the Indian system, however, the Ethiopian ar- lagging behind in service provision, the government of rangement measures the estimated tax base against Uganda first adopted an Equalization Grant in 1999. an assessment of the expenditure needs required for However, the formula adopted is at best weakly re- regional governments to meet basic service provision distributive – overall population size accounts for targets in assigned areas, including education, public over 90 percent of the weighting used in the formula health, agriculture and rural development, and roads governing allocations (Government of Uganda 2010). (Gebregziabher, Woldehanna and Ayenew 2012).

financing for a fairer , more prosperous kenya 59 Box 6: Federal Allocations to Regional States in levied by regional governments. The second mechanism Ethiopia is an Expenditure Needs Assessment, covering areas in which regional governments have assigned responsibili- Like Kenya, Ethiopia is marked by extreme regional dispari- ties. Education is the single largest regional budget item, ties in income and wider human development indicators. accounting for around one-third of the total, followed by (in Narrowing these disparities through intergovernmental descending order) administration, agriculture and rural de- transfers is a major priority. The grant formula developed velopment, health and water. The fiscal gap between each to determine allocations considers both the fiscal capacity region’s revenue raising capacity and expenditure need of regions and the estimated cost of achieving specified provides the basis for determining the level of transfers. goals – a framework that may have some relevance for Kenya. There are a number of equitable financing provisions built into the FBSA formula. The allocation for administrative Regional disparities in Ethiopia intersect with wider in - costs includes an equal share provision, with a 10 per - equalities linked to wealth, the rural-urban divide and cent supplement for hilly terrain and a higher per capita gender inequalities. While over 90 percent of the popula- transfer for pastoralist populations. Hardship allowances tion in Addis Ababa is in the country’s wealthiest quintile, averaging 30 percent are built into salary cost estimates that share falls to just 12 percent for Benishangul-Gumuz. for staff working in remote areas. In education, financing School attendance ratios range from 84 percent in Addis requirements are estimated on the basis of the per capita Ababa to 57 percent in the Somali region. Similarly, vacci- funding required to achieve the national education sector nation coverage ranges from 78 percent in Addis Ababa, to strategy target of full universal primary schooling. Because 15 percent in Gambella and 8 percent in Afar. These figures the formula takes into account the gap between current illustrate part of the rationale for intergovernmental trans- enrollment levels and target levels, it includes an implicit fers: regions with a large proportion of low-income house- premium in favor of regions with large out-of-school popu- holds (and hence a smaller tax base) face large deficits in lations. Similar approaches are applied to water and roads. the provision of basic services. In the case of health, financing costs weight for the number of people in each region living below the poverty line. The principle public spending mechanism for mitigating re- gional inequalities in Ethiopia is the Federal Budget Subsidy Looking beyond the allocation formula, Ethiopia has devel- Allocation (FBSA). Governed by a spending allocation for- oped a distinctive political process for determining alloca- mula that is subject to approval by parliament, the FBSA tions. The House of Federation in the Ethiopian parliament has to comply with constitutional provisions. These range effectively outsources technical work on the development from an obligation to ensure that “all Ethiopians get equal of the formula to high-level external consultants who work opportunity to improve their economic conditions”, the closely with staff from ministries, government agencies, promotion of an “equitable distribution of wealth” and eq- regional bodies, and revenue departments. Considerable uitable access to basic services, and the provision of “spe- emphasis is placed on participation and the development cial assistance” to the least advantaged counties. of a consensus across stakeholders before proposals are submitted to parliament. Regional allocations are determined through a grant for- Sources: Kenya National Bureau of Statistics 2010; mula covering two key areas. The first is an estimate of Gebregziabher, Woldehanna and Ayenew 2012. potential revenue, taking into account a wide range of taxes

Another problem is the size of the grant. As the gov- small scale of Kenya’s own Equalization Fund —0.5 ernment of Uganda has acknowledged: “The grant percent of national revenue—Uganda’s experience re- is too small to cover the vertical fiscal gaps and the inforces the case for an approach to equitable sharing differences in expenditure needs across local govern- that extends across all budget lines. ment” (Government of Uganda 2003). Given the very

60 Global Economy and Development Program The Ugandan evidence raises another question of and ‘equality’. The ‘equality’ provision relates solely relevance to the debate over equitable sharing of to population size and accounted for a 60 percent public spending in Kenya. That question is whether weighting of disbursements in 2005. ‘Need’ accounted the emphasis should be on equal spending across all for another 35 percent of the weighting, with the people, or on adjusting budget allocations to coun- formula combining six indicators for basic services. teract specific disadvantages. In the case of Uganda, However, the needs formulae have been changed on the most disadvantaged regions are located in the a regular basis, and the data needed to translate the north of the country—a legacy of several decades of formulae into allocations is often lacking. The result: armed conflict. Measured in per capita terms, overall allocations have been heavily influenced by political central government transfers to these regions are affiliations and electoral cycles rather than needs- comparable to the national average. Yet the northern based financing (Banful 2009). counties have deeper poverty, far lower levels of ba- sic service provision and face higher costs of service delivery, raising the question as to whether equi- Social Protection and Safety Nets table financing requires that they should be receiv- Intergovernmental transfers are just one weapon in ing higher per capita transfers (Uganda Multi-Donor the armory for correcting horizontal inequalities. In Group 2007). In the case of Kenya, as we show in the countries where income poverty is highly concen- next section, the Commission for Revenue Allocation trated in specific regions, targeted anti-poverty pro- has tended towards an ‘equal spending’ interpreta- grams will direct resources towards those regions. The tion of equitable sharing, rather than a needs-based same is true for programs targeting other forms of de- interpretation. Such an approach will weaken the role privation. In many cases programs targeting vertical of public spending in achieving the social goals set inequalities between people will have strong mutually- out in the 2010 constitution, including the reduction reinforcing effects and benefits for equity. of economic disparities between counties, support for marginalized areas, and moves towards the equaliza- The Bolsa Familia program in Brazil is an example. This tion of basic services. transfers around 0.5 percent of GDP to around 12 mil- lion households eligible by virtue of low income. On One of the clear lessons to emerge from a number one estimate, the program has accounted for around of countries is that complexity is best avoided. There one-sixth of the poverty reduction achieved in Brazil are strong political and financial grounds for adopt- since 2003. Independent evaluations have found posi- ing formulae that are transparent, easily understood tive results in terms of health and nutrition effects, and amenable to communication. Local government cognitive development, school enrollment, learning financing in Ghana has suffered from the absence of achievement and reductions in child labor. For ex - these simple virtues. The constitution provides for a ample, Bolsa Familia led to a 4.4 percent increase in ‘District Assemblies Common Fund’ representing at school attendance, with the largest gains occurring least 5 percent of national revenue to be distributed in the disadvantaged northeast, where enrollments through a formula agreed by parliament. The formula increased by 12 percent. Children covered by the Bolsa is supposed to reflect four underlying principles: Familia were also far more likely to progress from one namely, ‘need’, ‘responsiveness’, ‘service pressures’ grade to the next. This was especially true of girls

financing for a fairer , more prosperous kenya 61 aged between 15 and 17—who are at greatest risk of est safety net programs—budget allocations in 2009 dropping out. Bolsa Familia increased the likelihood reached $8.9bn—the scheme has generated rural em- that a 15-year-old girl will remain in school by 19 per- ployment, improved nutrition, and supported efforts centage points (de Brauw et al 2012). to raise school enrollment (International Food Policy Research Institute 2011). Ethiopia’s Productive Safety An analogous program in Mexico—Opportunidadas— Net Program (PNSP), sub-Saharan Africa’s largest uses geographical targeting and proxy-means testing social protection scheme outside of South Africa, is to make payments to over 4 million eligible house- modeled on the Indian program. In 2008 it reached 7 holds, with transfers given to the female family head. million people and had an operating budget of $500 Here, too, evaluations have found gains extending million. While the PNSP is not an education sector beyond the reduction of income poverty to improve- intervention, it has generated benefits in education, ments in child nutrition, cognitive development and including the reduced risk of children dropping out of progression through school (Behrman, Parker and school during drought episodes (Gilligan, Hoddinott, Todd 2008; Behrman and Hoddinott 2005). and Taffesse 2008).

Other social protection programs target vulnerable Social protection and safety-net programs have the po- populations without recourse to conditional trans- tential to simultaneously reduce horizontal disparities fers. In India the Mahatma Gandhi Rural Employment and enhance the self-reliance of vulnerable popula- Guarantee Program has provided a safety net for tions. They have a proven track record across many some 50 million households, offering guaranteed em- countries, although there are ongoing debates over ployment for 100 days a year and payment in cash or eligibility criteria, targeting mechanisms and the use food during periods of stress. One of the world’s larg- of conditional versus unconditional transfers (Fiszbein

Box 7: Vietnam’s Targeted Programs In parallel, Vietnam has a national program for Hunger Eradication and Poverty Reduction that targets house- Confronted with evidence of widening inequalities in health holds, rather than communes. In the education sector, the and slow progress among hard-to-reach groups in educa- government incorporated in its national strategy a Primary tion, Vietnam has adopted a number of ambitious pro- Education for Disadvantaged Children project aimed at grams. While the context is very different, the principles improving access to schools and raising learning standards underpinning the programs have a strong resonance with in over 400 districts characterized by low enrollment, the equity framework enshrined in Kenya’s constitution. high dropout rates and low levels of learning achievement. Expenditure was geared towards a range of inputs aimed Health disparities have been an area of concern in Vietnam. at raising the ‘fundamental school quality level’ in districts In 2002 the government introduced the Health Care Fund characterized by high levels of rural poverty and concen- for the Poor to finance free health care services, using trations of ethnic minority groups. Both the health and a mixture of group and geographic targeting. The fund education programs achieved significant results – and both now covers around 18 million people. Beneficiaries include demonstrate the impact of redistributive public spending households classified as poor according to the interna- in terms of expanding opportunity, narrowing gaps in basic tional poverty line, ethnic minority residents in six northern service provision, and cutting social disparities. provinces and five highland provinces, and all residents of 135 communes classified as socially disadvantaged. These Sources: Phuong 2009; Quan 2009. communes are also targeted through ‘Program 135’ for so- cial investments.

62 Global Economy and Development Program and Schady 2009). For the ASALs of Kenya, increased innovative technology-based systems for disbursing public spending on well-targeted social safety nets cash, including mobile transfers. Implementation is could help to weaken the transmission effects of in two phases. The principal objective of Phase 1 is to drought on poverty, child malnutrition and nonatten- implement a cash transfer program in the arid and dance at school. semi-arid lands districts of northern Kenya, making regular cash transfers to 69,000 households every While Kenya has a highly fragmented patchwork of two months for three years. Phase 2, beginning in safety nets offering limited coverage, recent years 2013, aims to roll out the HSNP under a national social have seen some progress towards the development protection system addressing the needs of 1.5 million of more comprehensive systems. The Hunger Safety Kenyans (some 400,000 households), with govern- Net Program (HSNP), an unconditional cash transfer ment of Kenya and donor funding. While these safety program targeted at the chronically food insecure, nets are not directly associated with local government has the goal of delivering long-term, guaranteed financing, they have consequences for equity in public cash transfers to the poorest and most vulnerable spending—and there are strong grounds for scaled-up 10 percent of Kenyan households. The project uses financing in support of the equitable sharing principle.

financing for a fairer , more prosperous kenya 63 Equitable Sharing for While much of the debate on equitable sharing in Kenya: The Education Sector Kenya has tended to focus on decentralized financ- and Beyond ing, the new constitutional principles apply to all national budgeting. Under the new devolved regime, s highlighted earlier in this paper, the Kenyan not less than 15 percent of all national revenue will Aconstitution establishes equitable sharing as a be allocated to devolved governments. There will also core value for public spending, but it does not provide be an Equalization Fund equivalent to 0.5 percent of guidelines on specific targets, the weight to be at - national revenue geared towards bringing the provi- tached to different aspects of equity, or the time-hori- sion of basic services in marginalized areas “to the zon over which opportunities – or outcomes – should level generally enjoyed by the rest of the nation.” The be equalized. These are issues that are the center of future of existing decentralized funds remains uncer- a national debate in Kenya. While the Commission tain. However, these arrangements will leave around on Revenue Allocation has produced some tentative 85 percent of revenue in the hands of central govern- ideas, that debate is likely to continue for some time ment and lines ministries – and these allocations will as the new system evolves. also have to be brought into line with constitutional principles. Designing a system geared towards more equitable sharing confronts policymakers with challenges at a This section looks at some of the issues that will have number of levels. Given the extreme nature of cross to be addressed in developing the formulae for al- county disparity in Kenya, it is evident that equitable locating resources and enacting constitutional prin- sharing has to mean something more than providing ciples. It starts by establishing as a point of reference equal amounts of finance for every citizen. Moreover, some initial proposals framed by the Commission on the constitution requires that public spending be Revenue Allocation, before looking at experience un- used to narrow disparities in access to basic services der existing devolved funds. We then consider what for marginalized groups and regions. This is a clear an approach to equitable financing might look like injunction to adopt a redistributive approach, even in education. Having developed some rule-of-thumb though the degree of redistribution is open to debate estimates for the current distribution of public spend- and to political negotiations. ing in education across counties, we look at a range of criteria that could be applied to allocate resources This starting point provides little guidance on the against need and we consider the cross county dis- weighting to be attached either to different dimen- tributional implications. While education will not be sions of inequality, or to current disparities in access subject to decentralization, the exercise serves to il- to basic services. Even if there were agreement in lustrate some wider equity issues. principle in these areas, the availability of data poses another layer of difficulties. Kenya lacks real-time data on deprivation for a wide range of indicators, An Initial Framework: The including poverty. Moreover, there are a range of un- Commission on Revenue Allocation resolved data issues over the population size in differ- Initial recommendations in February 2012 from the ent counties. Commission on Revenue Allocation (CRA) have

64 Global Economy and Development Program identified a number of key parameters for more equi- proposed by the CRA would provide the same level of table financing. The selection of parameters is instruc- per capita financing irrespective of how far people are tive, not least in highlighting some of the difficult issues from the poverty line. It is insensitive to the depth of that have to be addressed. Briefly summarized, the poverty. From an equitable financing perspective this CRA has focused on five core indicators of equity, with is problematic. If the aim is to align poverty-related a range of weights attached to reflect equity concerns. financing with poverty-related disadvantage, a better While these relate in the CRA’s recommendations to starting point for revenue allocation formula is the devolved budgets, the issues at stake have a wider rel- share of each county in the national poverty gap. One evance to Kenya’s constitutional requirements in favor advantage of a poverty gap approach is that it has a of equitable spending. The indicators are as follows: direct relevance for policy since it measures the mini- mum cost for eliminating poverty through transfers. • Population size (60 percent) to reflect average per An obvious limitation is that the poverty gap does not capita costs of service delivery and administration. fully capture the degree of inequality among the poor. • Equal shares across county (20 percent) to reflect While there are other technical tools that address this fixed county-level administration costs. shortcoming, including the squared poverty gap, Kenya

• Poverty (12 percent) under a formula that would may currently lack the data-base for their adoption. treat every Kenyan below the poverty line equally, The third concern with the CRA formula relates to ar- irrespective of their location. eas of omission. Household poverty is one important measure of deprivation—and it should figure promi- • Land area (6 percent) to reflect the additional costs nently in any equitable financing formula. But other of service delivery in large counties. deprivation indicators are also important. These could • Fiscal responsibility (2 percent) to reflect fiscal dis- include indicators for service availability (such as the cipline and create incentives for revenue mobiliza- ratio of doctors or nurses-to-population, immuniza- tion. tion rates or average distance to facilities), health sta- tus (as measured by child mortality, nutrition levels or Assessed as a mechanism for promoting equity, the the maternal mortality rate) and education (for exam- CRA formula suffers from a number of shortcomings. ple, the number of children out of school, gender gaps, Three problems stand out. The first is alimited weight- transition to secondary school or learning outcomes). ing for disadvantage. Under the proposed framework, 80 percent of the budget will be allocated without ref- Over and above these substantive points, there are erence to need on the basis of population and equal a number of technical difficulties with the proposed share provisions. It is geared towards an approach formula. Consider the provision for land area. This that implicitly views equitable sharing as a matter of is an attempt to address a real concern: namely, the providing equal transfers, irrespective of the balance higher unit cost of service provision in larger coun- of advantage and disadvantage across counties. ties. However, these costs may be more accurately reflected by population density than land area. In the Second, a related concern is a weak weighting for pov- case of the fiscal responsibility, it is not entirely clear erty and associated failure to consider the depth of which benchmarks will apply, what incentives will poverty. By definition, everyone below the poverty line emerge, or which counties stand to benefit. An obvious shares something in common. But the funding formula

financing for a fairer , more prosperous kenya 65 concern is that counties with limited revenue raising true of the Local Authorities Transfer Fund – until now capacity may lose out. More generally, there would ap- the primary source of local government financing. The pear to be strong grounds for dispensing with the fis- financing formula for transfers has been determined cal responsibility provision altogether and transferring by three factors: population size (60 percent), a lump- the proposed allocation to the poverty component. sum payment and the relative weight of the urban population. The criteria favor (wealthier) urban areas, with no weighting for poverty, wider human develop- Currently Devolved Funds ment disadvantages, inequalities in service provision, Currently devolved funds operate on a limited scale or the cost of service delivery. in Kenya. Even so, they serve to highlight some of the issues raised by the proposed CRA framework and by The Constituency Development Fund, which is man- the wider debate over equity. aged outside of the central government budget process, has some built-in weighting for equity. Equitable financing has been a limited guide to budget Allocations were initially made on the basis of an allocations in devolved financing, which is especially equal amount for each constituency (Bagaka 2008).

Figure 18: Actual CDF Allocations (2009-2010) vs . Inversion of the Current Formula (46 counties)





Garissa Narok Wajir 2 Kajiado West Pokot Turkana Mandera Tana River Actual CDF Allocation (percent) 1 Lamu Isiolo Marsabit 0 0 1 2 3 4 5 6 Share of New CDF Allocation (percent)

Source: CDF 2009/2010 Budget. *New CDF Allocation: 75 percent share of national poverty gap and 25 percent equal share. Note: Samburu is excluded because data for CDF allocation for 2009-2010 is missing.

66 Global Economy and Development Program Figure 19: CDF Budget Allocation Including Weighting for Population Density (46 counties)







Wajir 2 Narok Kajiado Garissa West Pokot Turkana Actual CDF Allocation (percent) Tana River Mandera 1 Lamu Isiolo Marsabit 0 0 1 2 3 4 5 6 7 8 Share of New CDF Allocation (percent)

Source: CDF 2009/2010 Budget. *New CDF Allocation: 50 percent share of national poverty gap, 25 percent equal, 25 percent counties with less than 40 people per km2 population density – divided based on each counties share of the national population. Note: Samburu is excluded because data for CDF allocation for 2009-2010 is missing.

This was subsequently amended so that 75 percent For understandable reasons, the CDF is popular with of the allocation is now determined on an equal share Kenyan parliament members and allocations have in basis, with the remainder distributed according to some cases made a significant contribution to local a formula that weights the constituency’s share in infrastructure. Viewed from an equity perspective, national poverty (Government of Kenya 2010). Apart however, the CDF is at best modestly redistributive in from the poverty criteria, the CDF has an implicit terms of its cross county effects. Despite their high weighting in favor of constituencies with smaller pop- levels of deprivation, the 12 ASAL counties receive a ulations (since this raises the per capita transfer from limited CDF preference. equal share financing). There is also a small weighting in favor of rural areas. The largest proportion of the For illustrative purposes, we compared current CDF CDF budget—just over one-third in recent years—has allocations to the ASAL counties with projected allo- been allocated to education. cations under two different formulae that attach more

financing for a fairer , more prosperous kenya 67 weight to identified disadvantages. Figure 18 con- introduction of free primary education in 2003 had a trasts the 2010 CDF allocation with the distribution strongly progressive effect, especially at the primary that would have occurred had transfers been based school level. With more children from the poorest on an inversion of the current formula: a 75 percent households entering the school system, the benefits allocation on the basis of share in national poverty of public spending in education have become more and 25 percent on the basis of equal resourcing. The widely distributed. However, existing education bud- outcome is mixed. Some of the 12 focus counties— get norms do little to redress the horizontal inequali- Turkana, Marsarbit, Isiolo and Mandera—gain, while in ties discussed in Section 2 of this paper. These norms others the effects are either neutral or slightly nega- strongly link transfers to the number of children in tive. Figure 19 includes a population-density weighting school, as distinct from the number of school-age chil- alongside the poverty and equal shares weighting. dren – an approach that effectively penalizes counties This produces a marked distributional shift in favor of with lower rates of enrollment, high levels of attrition most of the 12 ASAL counties covered in this report, in school progression, and lower levels of transition to although here too there are exceptions to the rule. secondary education.

These scenarios for the CDF illustrate an issue of The experience of the 12 ASAL counties illustrates wider relevance for approaches to equitable sharing. some of the wider equity problems. Partly because Some caveats have to be attached to our exercise they represent such a large share of Kenya’s out-of- because it is not clear that current CDF allocations re- school population in primary education and such a flect a strict application of the formula. However, if the small share of the secondary school-age population, aim is to develop a formula for decentralized financing children in the these counties receive less per capita that narrows the gap in basic service provision be - than those in counties registering better education tween the ASAL counties and other areas, attaching indicators. On a reasonable interpretation of equitable more weight to poverty and population density may financing, these counties should be receiving more in be a useful starting point. order to counteract the effects of household poverty, parental illiteracy, gender disadvantage, and wider so- cial and cultural barriers to equal opportunity. In this Towards Equitable Financing in section we provide an approximate estimate of the Education cross county pattern of education budget allocation Education will be the only major basic service sector and consider a range of alternative allocation formu- that is not subject to devolution. The sector accounts lae that might enhance equity. for around one-fifth of total government spending, or around 7 percent of GDP (Government of Kenya 2010). It follows that the rules governing education budget The Education Budget allocations will have a major bearing on equity in pub- Public spending on education has increased rapidly, ris- lic spending. ing by more than one-third in real terms over the past de- cade. Donors account for a relatively small share of the Existing budget arrangements have a mixed record overall budget. According to the 2010 Public Expenditure on equity. The surge in enrollment that followed the Review, aid accounted for 4 percent of overall

68 Global Economy and Development Program expenditure on education (Government of Kenya 2010). est recurrent expenditure items after teacher salaries Alongside public spending, households in Kenya make are the per capita pupil grants provided for both free significant out-of-pocket payments for education. primary education and free secondary education. These include expenditure on uniforms and learning materials, payments for private education provision Evolving budget expenditure patterns have significant and a range of formal and informal payments to public implications for the vertical and horizontal inequali- providers. ties in Kenya’s education system. With the share of secondary education in the overall budget envelope The distribution of benefits from public spending in rising over time, overall benefit incidence will in- education is a function of budget allocations and the creasingly reflect the profile of the secondary school pattern of school participation. Broadly, the closer a population. Socio-economic groups and counties with country is to universal primary and secondary educa- lower rates of school participation in the secondary tion the more equally distributed the benefits. Wealth- sector stand to lose out. Similarly, the distribution related benefit incidence analysis for Kenya carried of benefits from the per capita pupil grants for free out on the basis of 2005 data found that the poorest primary and secondary education will mirror the dis- 40 percent of children accounted for around half of tribution of children in school. Counties with the low- the benefits from government spending on primary est levels of enrollment will lose out relative to those education (children from wealthier households being with higher levels of enrollment. The same holds true more heavily represented in private schools). For sec- for the wider system of recurrent and capital funding. ondary education the share dropped to 15 percent, re- Current budget norms in Kenya allocate resources for flecting a low rate of transition from primary schools teachers, teaching materials and school infrastructure (Demery and Gaddis 2009). almost entirely on the basis of numbers of children enrolled in classrooms. The increase in public spending in Kenya has gone hand-in-hand with changes in the profile of budget In the following section we analyze budget allocation allocations. Primary education accounts for around through the prism of horizontal disparities and the one-half of overall public spending, but the share of position of the 12 ASAL counties. It is worth emphasiz- secondary education has increased from 20 percent ing, however, that vertical disparities intersect with to 24 percent since 2003. Per pupil expenditure at the what might be thought of as subcounty horizontal secondary level doubled in real terms between 2003- inequalities. For example, the limited provision of 2004 and 2008-2009, and has continued to increase public education in Kenya’s informal urban settle- with the introduction of free secondary education in ments forces many children out of the public educa- 2008. Around 60 percent of overall education spend- tion systems and into low-fee private schools. Given ing is accounted for by teacher salaries. Transfers in that these children are from households at the lower the form of grants and subsidies account for another end of Kenya’s income distribution, this skews public 30 percent. These transfers comprise payments as- spending for basic education in a less pro-poor direc- sociated with free primary and secondary education tion (Oketch and Ngware 2010; Oketch et al 2010). financing, grants to schools, bursaries and payments to local government for school supplies. The two larg-

financing for a fairer , more prosperous kenya 69 The 12 ASAL Counties ties are receiving a share of the budget allocation smaller than their share in the school-age population. Constructing an accurate benefit incidence curve In the case of Turkana, the budget share is less than that captures horizontal inequality in benefits inci- 40 percent of the county’s share in the school-age dence for Kenya would require disaggregated data on population. transfers of education financing by county. Data con- straints make it impossible to conduct this exercise. Secondary education allocations reflect the same pat- However, the close alignment of financial transfers tern in magnified form. High dropout rates in primary with school participation in primary education makes education and low rates of transition to secondary it possible to generate an approximation of the more school mean that the ASAL counties secure a small detailed picture. We derive an estimate of the primary share of the expanding secondary education budget. school budget allocations for the 47 counties from the Turkana is the national outlier with a budget share data provided in Section 2 of this paper. For second- equivalent to less than one-third of the county’s share ary education we estimate the per capita free primary in the secondary school-age population. Overall, the education grant by county, using this as a proxy for budget share of seven of the ASAL counties is less the overall budget allocation. Drawing on national than half of the school population share (Figure 21). census data, we then compare the county share in the With the exception of Kajiado, the 12 ASAL counties education budget with the county share of school-age fare badly in the current budget arrangements. They population. account for 11 of the 13 counties with the largest gap between secondary school population share and The results for primary education are presented in budget share. Several other counties also experience Figure 20. This shows that all 12 of the ASAL coun-

Figure 20: Derived County-Level Share of Primary Education Spending as a Proportion of School-Age Population (47 counties, 2009) 1.4 Less than primary More than primary school population share 1.2 school population share 1.0





0.0 Kisii Kilifi Kitui Nyeri Meru Wajir Wajir Siaya Isiolo Busia Embu Nandi Lamu Vihiga Narok Migori Bomet Kwale Nairobi Nakuru Kisumu Kericho Kiambu Kajiado Baringo Laikipia Garissa Turkana Turkana Nyamira Makueni Marsabit Mandera Kirinyaga Samburu Murang'a Bungoma Mombasa Homa Bay Machakos Kakamega Nyandarua Tana River Tana West Pokot Pokot West Trans Nzoia Trans Taita Taveta Taveta Taita Uasin Gishu Tharaka Nithi Elgeyo Marakwet

Source: Derived allocations based on school enrollment data and census data on school-age population.

70 Global Economy and Development Program Figure 21: Derived Share of FSE Spending as a Proportion of School-Age Population (47 counties, 2009)

1.8 1.6 More than the secondary school population share 1.4 Less than the secondary school population share 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Kisii Kilifi Kitui Nyeri Meru Siaya Wajir Wajir Busia Embu Nandi Isiolo Lamu Vihiga Migori Kwale Narok Bomet Nairobi Nakuru Kisumu Kericho Kiambu Baringo Laikipia Kajiado Garissa Nyamira Makueni Turkana Turkana Kirinyaga Mandera Marsabit Murang'a Bungoma Samburu Mombasa Homa Bay Machakos Kakamega Nyandarua Tana River Tana Trans Nzoia Trans Taita Taveta Taveta Taita Uasin Gishu West Pokot Pokot West Tharaka Nithi Elgeyo Marakwet

Source: Calculated on the basis of school enrollment data and census data on school-age population.

large gaps—depicted on the left-hand side of Figure of what is required under the equitable sharing provi- 21. The converse of this deficit is the ‘surplus’ shown sions on public spending enshrined in the 2010 con- on the right-hand portion of the figure, with Nairobi stitution. securing a budget share some 50 percent higher than the county’s population share. What might a more equitable set of budget norms look like? While there is no blueprint, two broad principles Even allowing for data uncertainties, the cross county would appear to be important. The first is that the pattern of budget allocation raises some fundamen- formula for budget transfers should be recalibrated tal questions about equity. Facing some of the most to attach more weight to provisions for school-age highly concentrated education disadvantages in children. Applying South Africa’s Provincial Equitable Kenya, children in the ASAL counties are receiving the Share formula for education illustrates one possible smallest share of the budget. Put differently, there is approach that may be of relevance for Kenya. This at- an inverse correlation between education deprivation taches equal weight in allocating finance to the overall and budget allocation. While any linkage between bud- number of school-age children and to the number of get transfers and school participation will automati- children in school. Figure 22 captures the distribu- cally have distributional effects mirroring enrollment tional shift that would occur if this formula were to be patterns, the strength of that linkage in Kenya means applied to primary education in Kenya. With the ex- that public spending may be reinforcing, rather than ception of Kajiado, which is only marginally affected, mitigating, social disadvantage. This is the opposite each of the 12 ASAL counties gains a larger share of

financing for a fairer , more prosperous kenya 71 Figure 22: Estimated Primary Education Budget Allocations by County: Equal Weighting Attached to School-Age Population and Children in School 6




Mandera Narok 2 Kajiado West Pokot Garissa Wajir 1 Turkana Share of Current Primary Budget (percent) MarsabitTana River Isiolo Lamu Samburu 0 0 1 2 3 4 5 6 Share of New Allocation (percent)

*New PES Allocation: 50 percent primary school-age population and 50 percent equal share. Note: Samburu and Tana River data points are overlapping on the figure. Source: EMIS 2010; Census 2009.

the budget. In the case of Wajir and Turkana the share deprivation, including household poverty, ill-health, increases by a factor of three, reflecting their low parental illiteracy and gender disadvantage. rates of school participation. Designing budget formulae that weight for education The second broad principle for equitable sharing in disadvantage is not straightforward. In Kenya, as in other education is a greater weighting for specific disad- countries, the underlying sources of unequal opportu- vantages. Adopting the South Africa PES approach nity in access to schooling and learning achievement are to funding would help to equalize budget transfers imperfectly understood. There are also large gaps in the per child. While such an outcome would meet a nar- availability of robust data. However, the disadvantages row interpretation of enhanced equity, it would fail associated with poverty are well established—and the on other tests. If the aim is to equalize education op- national poverty gap data are robust (if dated). The hori- portunity, providing children from very unequal back- zontal inequalities between counties in primary school grounds with equivalent resources is not a credible enrollment, progression and transition to secondary starting point. In the case of the 12 ASAL counties, education are also well established, with the Education more equal opportunities will require budget transfers Management Information System providing a national that counteract the underlying sources of education data source (see Section 2 of this paper).

72 Global Economy and Development Program Drawing on these sources it is possible to develop a • Primary school-age children not in school: 20 per- range of scenarios that demonstrate the implications cent (share of the national out-of-school popula- of changing the formulae guiding education budget tion). allocations. For illustrative purposes, we provide two • Gender disparity: 5 percent (allocated to counties reinforced equity scenarios – one for primary educa- with a ratio of girls-to-boys in school of less than tion, and one for secondary education. 0.95 on the basis of their share in the total number of out-of-school girls). Figure 23 captures the distributional shift that would • Household poverty: 20 percent (allocated on the occur with the implementation of what can be thought basis of the county-level share in the national pov- of as a reinforced equity formula for education financ- erty gap). ing based on the following allocation criteria: • An ASAL special fund: 5 percent (allocated to arid • Children in primary school: 50 percent (equal per and pastoral counties facing acute disadvantage capita funding). in education in proportion to their share in the pri- mary school-age population).

Figure 23: Estimated County-Level Primary Budget Allocations: Current Position vs . A Reinforced Equity Scenario*






Narok Mandera 2 Kajiado West Pokot Wajir 1 Garissa Turkana Tana River Marsabit Isiolo Samburu Share of Current Primary Education Budget (percent) Lamu 0 0 1 2 3 4 5 6 7 Share of New Allocation (percent)

Source: EMIS 2010; Census 2009. *New Allocation: 50 percent primary school-age children in school, 20 percent primary school-age children out of school, 20 percent national poverty rate, 5 percent counties less than 0.95 female-male ratio and 5 percent for ASAL counties.

financing for a fairer , more prosperous kenya 73 Any change in the public spending formula for educa- Figure 24 shifts the focus to secondary schooling. It tion will create winners and losers. The winners from traces the distributional shift that would result from a the reinforced equity formula appear to the right of formula based on the following elements: the 45 degree line in Figure 23, with the distance from • Children in school: 50 percent. the line capturing the size of the gain. Counties such as Garissa, Mandera, Marsabit, Turkana and Wajir all • Children not in school: 30 percent. double their share of the primary school budget. • Gender equity: 10 percent (allocated equally to Counties to the left of the line lose out in terms of counties with secondary Gross Enrollment Rates for budget shares, but the aggregate gain for the 12 ASAL girls of less than 40 percent). counties is the product of relatively modest declines in most of the other counties. • An ASAL special fund: 10 percent (allocated on the basis of population).

Figure 24: Secondary Education Allocations Under an Equity-Based Financing Formula: Comparison with Current Allocations (47 counties) **





2 Mandera Kajiado Narok 1 Wajir West Pokot Garissa Turkana Samburu Isiolo Marsabit Lamu Tana River 0 Share of Current Secondary Education Budget (percent) 0 1 2 3 4 5 Share of New Allocation (percent)

Source: EMIS 2010; Census 2009. Note: Samburu and Tana River data points are overlapping on the figure. **New Allocation: 50 percent secondary school children in school, 30 percent secondary school children out of school, 10 per- cent Gross Enrollment Rates for girls less than 40 percent and 10 percent for ASAL counties. Nairobi is not included in the data. The county accounts for about 10 percent of the current secondary education budget and 6 percent of the new allocation.

74 Global Economy and Development Program As in the case of primary education the reinforced of fairer, more inclusive societies, expanded opportu- equity provision creates a significant distributional nity and accelerated economic growth. Greater equity shift in favor of the 12 ASAL counties. For example, the in public spending can mobilize more finance for the budget share of Marsabit increases from less than 0.5 most disadvantaged counties. Whether or not those percent of the current education budget to almost 2 resources produce more equitable outcomes in the 12 percent. Turkana’s share rises from less than 1 percent ASAL counties will depend on the effectiveness of the to 3 percent. policies to which they are linked.

While both of our scenarios are illustrative they raise While this issue extends beyond the scope of the cur- some practical questions. The counterpoint to any eq- rent paper, the pattern of disadvantage in the ASAL uitable sharing reform that provides a better deal for counties is strongly suggestive of some priority ar- the most disadvantaged counties is the accompany- eas for additional financing. The high level of child ing adjustment in other counties. Preventing severe malnutrition and parental illiteracy in the counties disruption to education provision in counties that sets many children on a course for failure before they stand to lose out is obviously critical. The options are enter school. Expanded early childhood programs limited. Government can either increase the size of and preschool classes can make a difference. One re- the budget envelope to maintain current levels of per cent randomized control evaluation of a program in capita spending, with the increment in allocations go- Mozambique found that participation in a preschool ing to the most disadvantaged counties. Alternatively, program was associated with an increase of 24 per- it can adopt a gradualist approach, phasing in a cent in the likelihood of school enrollment, an addi- formula for greater equity over time. In a growing tional 7 hours per work spent on school activities, and economy with the level of budget revenue rising, this significant improvements in cognitive development could prevent more equitable sharing from becoming and problem solving abilities (Martinez, Naudeau and a zero sum game: the resources available to ‘losers’ Pereira 2012). These are results that underline the po- would still grow, but more slowly than those of ‘win- tentially high returns from preschool interventions – ners’. Ultimately, the design of any reform scenario is and they merit serious consideration in Kenya. At the about finding the right balance. In the case of Kenya, same time, the ASAL counties would clearly benefit that balance will involve steering a middle course that from a more integrated approach to child health and moves the country away from the existing pattern of preschool provision. budget allocation in the direction of more equitable sharing. Demand-side financing would appear to be another priority area. While the free primary and secondary education policies have lowered the cost of public Equitable Financing is Not a school, education is only nominally free. Households Substitute for Effective and Equitable still incur primary school costs for uniforms and text- Policies books. In remote rural areas parents have little choice More equitable public spending in education is not an but to send children to boarding schools for second- end in itself. It is a means through which governments ary education, and free provision does not extend to can create an enabling environment for the creation the full costs of boarding and ancillary expenditures

financing for a fairer , more prosperous kenya 75 (Ohba 2009; Oketch and Somerset 2010). High levels Cambodia found strong effects on enrollment, es - of poverty in the ASAL counties mean that many par- pecially among girls from the poorest 20 percent of ents are unable to afford indirect primary school costs households (Filmer and Schady 2008; Fiszbein and and, even more so, secondary school fees. Moreover, Schady 2009). Here, too, there are potential guides social and cultural practices mean that when financial for policy design in northern Kenya. pressures force schooling choices, girls lose out rela- tive to boys. Investments on the supply-side of education provision could also make a difference. Low population densi- More equitable public spending in education could ties in many of the ASAL counties reduce the size of help to counter financial barriers to education. Cash schools and increase distance from the home, while transfers can be provided to the poorest households the remote nature of many districts makes it difficult through social protection schemes. One approach to attract and retain teachers. Increased spending on might be to finance an education facility within the innovative approaches to service delivery, including Hunger and Safety Net Program. This currently mobile schools, ‘satellite schools’ and distance-learn- reaches an estimated 450,000 food insecure people ing provision, could extend the reach of education sys- in four of the poorest counties – Turkana, Marsabit, tems. Teacher recruitment and retention poses wider Mandera, Wajir – and will gradually scale up to reach challenges. Even so, part of the revenue generated 2.5 million. Currently, Kenya under-invests in this area: through more equitable financing could be directed the country spends just 0.25 percent of GDP on social towards improved incentives and hardship allowances protection, with 80 percent of that amount directed for teachers. towards emergency support (World Bank 2011b). Not all of the problems faced by the 12 ASAL counties Some consideration could also be given to the de- can be addressed in isolation from the wider chal- velopment of conditional cash transfer programs, lenges facing the education system in Kenya. The with eligibility linked to the attendance of children in system is missing many of the ingredients found in school – and with a premium payment for girls. The countries that have successfully raised learning stan- effectiveness of such programs hinges on the target- dards. Kenya lacks a functioning learning assessment ing of support on the critical dropout years. While to identify failing schools and pupils. Teacher training these vary across the 12 ASAL counties, the transition programs are not equipping teachers with the skills from primary to secondary school is a major source they need to deliver effective learning, especially in of school attrition. Stipend and bursary programs the early grades. There is no policy framework linking can create further incentives for keeping children in financial resource allocation to learning outcomes. To school. In Bangladesh, secondary school stipends for a greater or lesser degree every county in Kenya suf- girls have been credited with the creation of incen- fers from these wider weaknesses, though the ASAL tives for households to ensure that their daughters counties covered in this report suffer disproportion- complete primary education (Khandker, Pitt and ately more than most. Fuwa 2003). Similarly, a stipend program for girls in

76 Global Economy and Development Program Conclusion equitable sharing provisions. The constitution itself identifies affirmative action to reduce inequality he new constitutional framework for public fi- and overcome marginalization as core priorities, Tnance in Kenya marks a bold departure from the suggesting that public spending allocations should norm. Translating the principles behind that frame- be positively associated with needs—that is, the work into operational practice will require strong greater the degree of disadvantage the higher the political leadership. Some of the poorest and most vul- level of per capita support that should be provided. nerable sections of Kenyan society stand to emerge as • The poverty gap, as distinct from the poverty winners – but this is not a constituency with a strong headcount, should be a primary indicator of need voice. The danger is that the process of political ne- in formulae governing allocations across coun- gotiation will see the potential gains for poor people ties, and in national budgeting for nondevolved and marginalized areas diluted. Guarding against that sectors. Initial proposals from the Commission for Revenue Allocation applied the ‘equal share’ prin- danger will require political figures of all persuasions ciple to poverty-related transfers, with an equal to forge a new consensus in favor of a more equal cash transfer for every person in a county below the society, backed by a shared commitment to support national poverty line. This is a flawed starting point more equitable public spending. because the proposed transfer would be unrelated to the depth of poverty. The ‘poverty gap’ is a more This paper sets out some broad approaches that sensitive measure of disadvantage. Poverty-related transfers should be based on the share of each could translate the commitment to equitable sharing county in the national poverty gap. of public spending into practical policies. It highlights the critical role of the formulae that will be used to de- • The poverty gap should have a significant termine financial allocations. The following are among weighting in the devolved budget formula. We rec- the core recommendations to emerge: ommend that consideration be given to a formula that attaches a weight of 30 to 50 percent to the • The constitution’s ‘equitable sharing’ provision share of each county in the national poverty gap, should apply to all public spending, and not just with additional weighting for low population density the devolved budgets. The distinction is impor- counties. While the poverty gap and population den- tant. Devolved financing will cover up to 15 percent sity are proxies for wider disadvantages, they are of national revenue, with a further 0.5 percent of indicative of both the level and intensity of depriva- revenue allocated through an Equalization Fund. tion, and of the costs of moving towards more equal However, the constitution requires that “all aspects opportunities on the other. of public finance…promote an equitable society” (Government of Kenya 2011a). Both the letter and • The budget allocation framework for primary the spirit of its public spending provision require education should place greater emphasis on the that government extends the ‘equitable sharing’ equalization of opportunity, with weighting at- principle to the 85 percent of the budget falling out- tached to the number of school-age children and side of the more narrowly-defined devolved funds. other indicators of deprivation. Current education financing norms allocate resources almost entirely • Proposals from the Commission for Revenue to reflect numbers of children in school. This has Allocation should attach less weight to equal per an unintended perverse effect in that it implicitly capita transfers and more weight to indicators penalizes counties with lower levels of school entry for disadvantage. Equal spending per capita repre- and higher dropout rates, including most of the 12 sents a minimalist interpretation of the constitution’s ASAL counties. Moreover, the current formulae for

financing for a fairer , more prosperous kenya 77 budget allocation attach no weight to wider indica- the additional resources created for counties being tors of disadvantage—such as household poverty left behind are used to finance policies that will or parental illiteracy—which influence the distri- promote greater equity in access to education and bution of opportunities in education. We therefore learning outcomes. These policies could range from advocate a financing approach that attaches more conditional (or unconditional) cash transfers to weight to (i) the total number of school-age children targeted financing for bursaries, support for girls, in a county (ii) the poverty gap and (iii) broader investment in teaching materials, or increased in- indicators of deprivation and inequality, including service support for teachers in ASAL counties. In gender disparities. One approach might be to con- the case of the 12 ASAL counties, the financing pre- sider weighting in the following ranges: 50 percent mium generated through more equitable sharing for children in school, 20 percent for children not could be geared towards the development of a more in school, 20 percent for household poverty, 5 per- responsive and relevant system, including support cent for gender disparity and 5 percent allocated for mobile learning facilities geared towards pasto- to a special fund for arid and pastoralist counties. ralist livelihood patterns, distance learning, and ad- Variations on this approach could be considered. aptation of the curriculum and teaching materials.

• Secondary education financing should be re- • Looking to the future, policymakers in Kenya viewed in the light of the acute disadvantages should consider the development of a public fi- facing the ASAL counties. The limited progression nancing model geared towards the provision of a of children from the 12 ASAL through Form 4 of sec- ‘social minimum’ of basic services. Through its na- ondary school and the KCSE is indicative of deeply tional commitments to the Millennium Development entrenched structures of disadvantage, especially Goals and the social and economic rights enshrined for young girls. Learning achievement levels also in the constitution, Kenya has committed to provide lag behind the national average. In developing a a basic standard of provision for all citizens. As the financing formula that reflects the constitutional devolved system develops over time, county-level principle of equitable sharing, policymakers should and national authorities should estimate the financ- consider linking half of the budget allocation to in- ing gap facing each county with respect to the pro- dicators for disadvantage. The scenario presented vision of key basic services. That gap should figure in this paper considers a 50 percent allocation as a ‘needs assessment’ component in national and geared to equal financing for children in school, devolved financing formulae. with 30 percent based on the number of children out of school, 10 percent on gender disadvantage • The government of Kenya and aid donors should (as indicated by a GER of less than 40 percent), and invest in building the national statistical capacity with 10 percent of the allocation reserved for ASAL required to underpin a devolved financing system counties. and to inform approaches to equitable sharing. Kenya has a relatively strong and professionalized • More equitable public spending in education set of institutions generating data on social and should be linked to the development of policies economic indicators, including the Kenya National and delivery mechanisms aimed at translating Bureau of Statistics and relevant line ministries. increased financing into more equitable opportu- There are problems however. Surveys on key human nities for access and learning in the ASAL coun- development indicators are intermittent. In some ties. The 12 ASAL counties covered in this report areas the available data is either dated or not avail- have some of the lowest levels of participation in able on a comparable basis across counties. In oth- education and some of the worst learning achieve- ers, the accuracy of the data available is contested. ment levels in Kenya. More equitable public spend- The bottom line is that policymakers currently lack ing could help to change this picture, provided that access to the reliable, real-time data required to

78 Global Economy and Development Program inform approaches to equity. If one of the criteria The provisions of the 2010 Kenyan constitution do for ‘equitable sharing’ is an allocation of resources not define a course towards a more equitable society. that reflects need, it is important to develop statis- What the constitution does provide is an injunction tical systems that capture relative deprivation in a to put inequality, marginalization and poverty at the timely and systematic fashion geared towards an- centre of the agenda for public spending. By exten- nual budgeting cycles. sion, this implies that the ASAL counties, which are There are limits to what can be achieved through pub- home to some of the most marginalized communi- lic spending. As even a casual glance at the budgets ties in Kenya, should get a better deal in future public of many developing and developed countries would spending rounds. Whether that outcome materializes confirm, high levels of spending on basic services do will ultimately depend on the degree to which political not automatically translate either into higher levels of leaders are guided by the spirit and the letter of the human development, or into expanded coverage and new constitution. improved quality. Outcomes in these areas are deter- mined by the efficiency of public spending and the level of equity in budget allocations.

financing for a fairer , more prosperous kenya 79 References 2008. [Accessed from http://mpra.ub.uni-].

Adams, S. (2005) Vietnam’s Health Care System: A Bahl, R. (2008) The Pillars of Fiscal Decentralization. Macroeconomic Perspective. Paper presented CAF Working Papers, No. 2008/07. [Accessed from at the International Symposium on Health Care]. Systems in Asia. Tokyo, 21-22 January 2005. Hanoi: International Monetary Fund. Bahl, R. (2010) Financing Subnational Governments with Decentralized Taxes. No. 1020, Working Africa Progress Panel (2012) Africa Progress Report Paper Series. International Center for Public 2012: Jobs, Justice and Equity-Seizing opportu- Policy, Andrew Young School of Policy Studies. nities in times of global change. Geneva: Africa Atlanta: Georgia State University. Progress Panel. Bahl, R. and Linn, J. (1994) Fiscal Decentralization and Akech, M. (2010) Institutional Reform in the New Intergovernmental Transfers in Less Developed . Nairobi: International Countries. Publius: The Journal of Federalism. Center for Transitional Justice. 24(1). 1-19.

Alesina, A. and Rodrik, D. (1994) Distributive Policies Bahl, R. and Wallace, S. (2004) Intergovernmental and Economic Growth. The Quarterly Journal of Transfers: The Vertical Sharing Dimension. No. Economic Growth. 109(2) May 1994. 465-490. 0419, Working Paper Series. International Center for Public Policy, Andrew Young School of Policy Alm, J. and Martinez-Vazquez, J. (2009) South Africa’s Studies. Atlanta: Georgia State University. Provincial Equitable Share: An Assessment of Issues and Proposals for Reform. No. 0904, Banful, A.B. (2009) Can Institutions Reduce Working Paper Series. International Center for Clientelism? A Study of the District Assemblies Public Policy, Andrew Young School of Policy Common Fund in Ghana. IFPRI Discussion Paper Studies. Atlanta: Georgia State University. 0085, April 2009. Washington D.C.: International Food Policy Research Institute (IFPRI). Axelson, H., Bales, S., Minh, P., Ekman, B and Gerdtham, U. (2009) Health financing for the Behrman, J.R., and Hoddinott, J. (2005) Program eval- poor produces promising short-term effects on uation with unobserved heterogeneity and selec- utilization and out-of-pocket expenditure: evi- tive implementation: The Mexican PROGRESA dence from Vietnam. International Journal for impact on child nutrition. Oxford Bulletin of Equity in Health, 8(20). Economics and Statistics. 67(4). 547-569.

Bagaka, O. (2008) Fiscal Decentralization in Kenya: Behrman, J.R., Parker, S.W., and Todd P.E. (2009) The Constituency Development Fund and the Schooling Impacts of Conditional Cash Transfers Growth of Government. MPRA Paper No. 11813. on Young Children: Evidence from Mexico. Published in the Proceedings of 20th Annual Economic Development and Cultural Change. Conference of the Association for Budgeting and April 2009. 57(3). 439-477. Financial Management. Chicago, 23-25 October

80 Global Economy and Development Program Berg, A. and Ostry, J. (2011) Inequality and Bold, T., Kimenyi, M., Mwabu, G., and Sandefur, J. Unsustainable Growth: Two Sides of the Same (2011a) Why Did Abolishing Fees Not Increase Coin? IMF Staff Discussion Note No. 11/08. Public School Enrollment in Kenya? Working Washington D.C.: International Monetary Fund. Paper 271. Washington D.C.: Center for Global Development. Bergh, G. and Melamed, C. (2012) Inclusive growth and a post-2015 framework. London: Overseas Bold, T., Kimenyi, M., Mwabu, G. and Sandefur, J. Development Institute. (2011b) The High Return to Private Schooling in a Low-Income Country. Working Paper 279. Bermingham, D. (2011) The Interactions Between Global Washington D.C.: Center for Global Development. Education Initiatives and National Education Policy and Planning Processes: A Comparative Bourguignon, F., Ferreira, F., and Walton, M. (2007) Case Study of the Education For All Fast Track Equity, Efficiency and Inequality Traps: A Initiative in Rwanda and Ethiopia. CREATE Research Agenda. Journal of Economic Inequality. Pathways to Access Research Monograph No. 67. (5). 235-256. Brighton: University of Sussex. Brown, G. (2011) Education for All: beating poverty, Bird, R.M., and Bahl, R. (2008) Subnational Taxes unlocking prosperity, a review by Gordon Brown in Developing Countries: The Way Forward. No. [Accessed from 16, Working Papers Series. Rotman Institute of brown/EFA%20Report_lores_FIN.pdf]. International Business. Toronto: University of Toronto. Bruns, B., Evans, D., and Luque J. (2012) Achieving World Class Education in Brazil: The Next Agenda. Birdsall, N. (2007) Income Distribution: Effects on Washington D.C.: The World Bank. Growth and Development. Center for Global Development Working Paper No. 118, April 2007. Chakraborty, P. (2010a) Deficit Fundamentalism vs. th Washington D.C.: Center for Global Development. Fiscal Federalism: Implications of 13 Finance Commission’s Recommendations. Economic & Boex, J. and Kelly, R. (2011) Fiscal Decentralization in Political Weekly. Vol XLV No 48. November 27, Kenya: A Small Step of Giant Leap? IDG Policy 2010. 56-63. Brief. Washington D.C.: Urban Institute Center on th International Development and Governance. Chakraborty, P. (2010b) Report of the 13 Finance Commission: Introduction and Overview. Boex, J. and Martinez-Vazquez, J. (2005) The Economic & Political Weekly. Vol XLV No 48. 27 Determinants of the Incidence of Intergovernmental November 2010. 35-37. Grants: A Survey of the International Experience. No. 0509, Working Paper Series. International Chowdry, H. and Sibieta, L. (2011) School funding re- Studies Program, Andrew Young School of Policy form: an empirical analysis of options for a na- Studies. Atlanta: Georgia State University. tional funding formula. IFS Briefing Note BN123. London: Institute for Fiscal Studies.

financing for a fairer , more prosperous kenya 81 Commission on Revenue Allocation (2011) Kenya: Economic Commission for Latin American and the Country Fact Sheets December 2011. Nairobi: Caribbean (ECLAC) (2011) Social Panorama of Commission on Revenue Allocation. Latin America: 2010. Santiago: United Nations.

Commission on Revenue Allocation (2012) Promoting Eurydice at The National Foundation for Educational an equitable society: Formula For Sharing Research (NFER) (2001) The Financing of Schools Revenue Among County Governments. Power in England, Wales and Northern Ireland. [Accessed Point Presentation. 24 April 2012. Nairobi: from]. Commission on Revenue Allocation. Fengler, Wolfgang. “Imagine a new Indonesia: Commonwealth Secretariat (2007) Reaching nomadic Spending to improve development.” East Asia & populations in Africa-Summary Report. Forum on Pacific on the rise: Making development work for Flexible Education. Garissa, 20-23 June 2006. all (blog), World Bank. 4 June 2009. [Accessed 20 September 2011, from Cornia, G. (2012) Inequality Trends and their Determinants: eastasiapacific]. Latin America over 1990 to 2010. Working Paper No. 2012/09. Helsinki: United Nations University- Ferreira, F. (2001) Education for the Masses?: The World Institute for Development Economics Interaction between Wealth, Educational and Research (UNU-WIDER). Political Inequalities. Economics of Transition. 9(2). 533-552. De Brauw, A., Gilligan, D., Hoddinott, J., and Roy, S. (2012) The Impact of Bolsa Familia on Child, Ferreira F. and Gignoux, J. (2011) The Measurement Maternal and Household Welfare. Washington of Inequality of Opportunity: Theory and an ap- D.C.: International Food Policy Research Institute plication to Latin America. Review of Income and (IFPRI). Wealth. 57(4). 622-657.

Demery, L. and Gaddis, I. (2009) Social spending, Ferreira, F. and Ravallion, M. (2009) Global Poverty poverty and gender equality in Kenya: a ben- and Inequality: A review of the Evidence. Chapter efit incidence analysis. Support to Public Finance 24 in Oxford Handbook of Economic Inequality. Management Reforms. Nairobi: GTZ. Edited by W. Salverda, W. Nolan, and T. Smeeding. New York: Oxford University Press. Diderichsen, F. (2004) Resource Allocation for Health Equity: Issues and Methods. Health Nutrition and Filmer, D. and Schady, N. (2008) Getting Girls into Population (HNP) Discussion Paper. Washington School: Evidence from a Scholarship Program in D.C.: The World Bank. Cambodia. Economic Development and Cultural Change. 56(3). 581-617. Domingo, P. and Wild, L. (2012) Will Kenya’s 2010 Constitution work for women and children? Financial and Fiscal Commission (2009) Technical Project Briefing No. 74, May 2012. London: Report: Annual Submission on the Division Overseas Development Institute. of Revenue 2010/2011. Edited by T. Ajam, B. Khumalo, R. Mabugu and B. Setai. Johannesburg: Financial and Fiscal Commission.

82 Global Economy and Development Program Fiszbein, A. and Schady, N. (2009) Conditional Cash Government of Kenya (2008) Education Statistical Transfers Reducing Present and Future Poverty. Booklet 2003-2007. Nairobi: Ministry of World Bank Policy Research Report. Washington Education. D.C.: The World Bank Government of Kenya (2009a) Education Management Fitzgibbon, C. (2012) Economics of Resilience Study- Information System (EMIS). Nairobi: Ministry of Kenya Country Report. Nairobi: UNDP, (MIMEO). Education.

Fukuyama, F. (2012) Dealing with Inequality in Poverty, Government of Kenya (2009b) Kenya National Census. Inequality, and Democracy. Edited by F. Fukuyama, Nairobi: Kenya National Bureau of Statistics. L. Diamond, and M. Plattner. Baltimore: John Hopkins University Press. Government of Kenya (2010) Public Expenditure Review: Policy for Prosperity 2010. Nairobi: Ministry Gebregziabher, T., Woldehanna, T., and Ayenew, M. of State for Planning, National Development and (2012) Federal Democratic Republic of Ethiopia: Vision 2030. [Accessed from http://www.plan- Federal Budget Subsidy Allocation Formula to]. Regional States. Addis Ababa: Federal Democratic Republic of Ethiopia, (MIMEO). Government of Kenya (2011a) Devolved Government & Public Finance: Extracts from the Constitution Gilligan, D.O., Hoddinott, J., and Taffesse, A.S. (2008) of Kenya. Nairobi: Commission on Revenue The Impact of Ethiopia’s Productive Safety Net Allocation. Programme and Its Linkages. IFPRI Discussion Papers No. 839. Washington D.C.: International Government of Kenya (2011b) Education Facts and Figures. Food Policy Research Insitute (IFPRI). Ministry of Education. [Accessed from http://www.]. Glennerster, R., Kremer, M., Mbiti, I. and Takavarasha, K. (2011) Access and Quality inthe Kenyan Government of Uganda (1999) Introduction of Education System: A Review of the Progress, Equalization Grants for Districts: Analysis Challenges and Potential Solutions. Cambridge: and Recommendations No. 1. Kampala: Local MIT J-PAL. [Accessed from www.povertyaction- Government Finance Commission. [Accessed from]. education-system]. Government of Uganda (2003) Allocation Principles, Government of Kenya (2006) Kenya Integrated Formulae, Modalities and Flow of Central Household Budget Survey (KIHBS) 2005-06. Government Transfers-Phase One: Commission Nairobi: The World Bank. Recommendations No.9 2003. Kampala: Local Government Finance Commission. [Accessed Government of Kenya (2007) Kenya Vision 2030. from]. Nairobi: Government of Kenya.

financing for a fairer , more prosperous kenya 83 Government of Uganda (2010) Ministerial Policy Millennium Development Goal 5. The Lancet. 375. Statements: Presented to Parliament For the 1609-1623. Debate on Budget Estimates, Financial Year 2010- 2011. Kampala: Ministry of Local Government. International Food Policy Research Institute (2010) Conditional Cash Transfers in Latin America. Government of the United Kingdom (2011) The Local Edited by M. Adato and J. Hoddinott. Baltimore: Government Finance Report (England) 2011/2012. John Hopkins University Press. London: Department for Communities and Local Government. International Food Policy Research Institute (2011) Regional Developments: Food Policy Taking Shape Hanson, S. (2008) Understanding Kenya’s Politics. at the Local Level, Chapter 9 in 2011 Global Food Council on Foreign Relations (CFR) Backgrounder, Policy Report. Washington D.C.: International January 25, 2008. [Accessed from http://www. Food Policy Research Institute (IFPRI). 79-87. p15322]. International Monetary Fund (2010) Kenya: Poverty Reduction Strategy Paper (PRSP). IMF Country Hanushek, E. (2009) The Economic Value of Education Report No. 10/224. Washington D.C.: International and Cognitive Skills in Handbook of Education Monetary Fund. Policy Research. Edited by G. Sykes, B. Schneider and D. Plank. New York: Routledge. 39-56. Isaac, T. and Chakraborty, P. (2008) Intergovernmental Transfers: Disquieting Trends and the Thirteenth Henriques, R. (2009) Educational marginalization in Finance Commission. Economic & Political Brazil. Background paper for UNESCO EFA Global Weekly. Vol XLIII No.43, October 25, 2008. 86-92. Monitoring Report 2010. Kabubo-Mariara, J., Mwabu, D. and Ndeng’e, G. (2012) Higgins, K., Bird, K. and Harris, D. (2010) Policy re- Institutions, Pro-Poor Growth and Inequality sponses to the spatial dimensions of poverty. ODI in Kenya. Paper presented at the CSAE 2012 Working Paper 328/CPRC Working Paper 168. Conference on Economic Development in Africa., London: Overseas Development Institute. St. Catherine’s College. Oxford, 18-21 March 2012.

Hofman, B. and Kaiser, K. (2002) The Making of the Kenya National Bureau of Statistics (2009) Kenya: Big Bang and its Aftermath: A Political Economy 2009 Population and Housing Census Highlights. Perspective. Paper presented at the conference: Nairobi: Kenya National Bureau of Statistics. Can Decentralization Help Rebuild Indonesia? [Accessed from International Studies Program Conference, Results/KNBS%20Brochure.pdf]. Georgia State University. Atlanta 1-3 May 2002. Kenya National Bureau of Statistics (2010) Kenya Hogan, M., Foreman K., Naghavi, M., Ahn, S., Wang, Demographic and Health Survey (KDHS) 2008- M., Makela, S., Lopez, A., Lozano R, and Murray C. 2009. Nairobi: Kenya National Bureau of (2010) Maternal Mortality for 181 Countries, 1980- Statistics. 2008: A Systematic Analysis of Progress Towards

84 Global Economy and Development Program Kenya National Bureau of Statistics (2011) Sectoral Lefkowits, L. (2004) School Finance: From Equity Reports, Education Enrollment from Statistical to Adequacy. Policy Brief March 2004. Aurora, Abstract 2011 [Accessed 1 November 2011, from CO: Mid-Continent Research for Education and Learning (McREL). php]. Lieberman, S. (2002) Decentralization and Health Kenya National Commission on Human Rights (2011) in the Philippines and Indonesia: An Interim Making the Bill of Rights Operational: Policy, Legal, Report. East Asia and Pacific Regional Overview. and Administrative Priorities and Considerations. Washington D.C.: The World Bank. Occasional Report, October 2011. Nairobi: Kenya National Commission on Human Rights. Lieberman, S., Capuno, J. and Van Minh, H. (2005) Decentralizing Health: Lessons from Indonesia, Kenya National Dialogue and Reconciliation (2011) the Philippines, and Vietnam in East Asia Building A Progressive Kenya: Our Common Decentralizes: Making Local Government Work. Vision, Views of Stakeholders. Geneva: Kofi Annan Washington D.C.: The World Bank. Foundation. [Accessed from http://www.kndr.]. MacDonald, R. and Hallwood, P. (2004) The Economic Case for Fiscal Federalism in Scotland. The Khandker, S., Pitt, M. and Fuwa, N. (2003) Subsidy to Allander Series, University of Strathclyde Promote Girls’ Secondary Education: The Female [Accessed from Stipend Program in Bangladesh. MPRA Paper allanderseries/allanderseriespapers/] . No. 23688 [Accessed from http://mpra.ub.uni-]. Martinez, S., Naudeau, S., and Pereira, V. (2012) The Promise of Preschool in Africa: A Klasen, S. (1999) Does Gender Inequality Reduce RandomizedImpact Evaluation of Early Childhood Growth and Development? Evidence from Cross- Education in Rural Mozambique. The World Bank Country Regressions. World Bank Policy Research Group and Save the Children. Washington D.C.: Report Working Paper No.7. Washington DC: The The World Bank. World Bank. Martorell, R., Horta, B.L., Adair, L.S., Stein, A.D., Richter Kramon, E. and Posner, D. (2011) Kenya’s new consti- L., Fall, C.H.D., Bhargava, S.K., Dey BiswasS.K. , tution. Journal of Democracy. April 2011. 22(2) Perez, L., Barros, F.C., and Victoria, C.G. (2010) 89-103. Weight Gain in the First Two Years of Life is an Important Predictor of Schooling Outcomes in Krätli S. and Dyer, C. (2009) Mobile Pastoralists and Pooled Analyses From Five Birth Cohorts from Education: Strategic Options. Education for Low and Middle Income Countries. The Journal of Nomads Working Paper 1. London: International Nutrition. February 2010. 140(2). 348-354. Institute for Environment and Development. McLeod, F. and McLeod, R. (2010) Fiscal Transfers to Kumhof, M., and Ranciere, R. (2010) Inequality, Leverage Regional Governments in Indonesia. Working pa- and Crises. IMF Working Paper WP/10/268. per No. 2010/14. Canberra: Australian National Washington DC: International Monetary Fund. University.

financing for a fairer , more prosperous kenya 85 Melamed, C. (2012) Putting inequality in the post-2015 Ohba, A. (2009) Does free secondary education en- picture. London: Overseas Development Institute. able the poor to gain access? A study from rural Kenya. CREATE Pathways to Access Research Menon, B., Mutero, J., and Macharia, S. (2008) Monograph No. 21. Brighton: University of Sussex. Decentralization and Local Governments in Kenya. Working Paper 08-32. International Studies Oketch, M., Mutisya, M., Ngware, M., and Ezeh, A.C. Program, Andrew Young School of Policy Studies. (2010) Why are there proportionately more Atlanta: Georgia State University. poor pupils enrolled in non-state schools in ur- ban Kenya in spite of FPE policy? International Ministry of Education, Kenya (2010) Policy framework Journal of Educational Development. 30(1). 23-32. for nomadic education in Kenya. Nairobi: Ministry of Education. Oketch, M. and Ngware, M. (2010) Free primary educa- tion still excludes the poorest of the poor in urban Muhula, R. (2009) Horizontal Inequalities and Ethno- Kenya. Development in Practice. June 2010. 20(4- regional politics in Kenya. Kenya Studies Review 5). 603-610. (KSR). December 2009. 1(1). 85-105. Oketch, M. and Somerset, A. (2010) Free Primary Muindi, B. (2012a) Kenya: Academies Rig KCPE Education and After in Kenya: Enrollment and im- Rankings. Daily Nation. 12 January 2012. [Accessed pact, quality effects, and the transition to second- from ary school. CREATE Pathways to Access Research html]. Monograph No. 37. Brighton: University of Sussex.

Muindi, B. (2012b) Kenya: Bulk of National Slots Ozier, O. (2010) The Impact of Secondary Schooling Go to Public Schools. Daily Nation, 12 January Kenya: A Regression Discontinuity Analysis. 2012. [Accessed from Berkeley: University of California ries/201201131321.html]. Petchey, J., MacDonald, G., Josie, J., Mabugu, R. and Murray, C. (2011) The Constitution of Kenya and the Devolved Kallis, D. (2007) A Grant Scheme for the Progressive Government. [Accessed November 2011 from, Realisation of Constitutionally Mandated Basic Services in South Africa: A Simulation Model. content&view=article&id=49:the-constitution-of- Midrand: Financial and Fiscal Commission. kenya-and-the-devolved-government&catid=34:lib rary&Itemid=61]. Pham, L., Minh, T., and Duong, N. (2006) Impacts of Public Spending on Growth and Poverty Alleviation Ngware, M., Oketch, M., and Ezeh, A.C. (2011) Quality in Vietnam: A Dynamic General Equilibrium of Primary Education Inputs in Urban Schools: Analysis. MPIA Network Session Paper. Paper pre- Evidence From Nairobi. Education and Urban sented at the 5th PEP Research Network General Society. 43(1). 91-116. Meeting. Addis Ababa, 18-22 June 2006.

Nyagesiba, B. (2012) 30 Private Schools in Kisii Closed Phuong, N. (2009) Viet Nam: Review of Financing of Indefinitely. The Star, 30 May 2012. [Accessed from Health-care, including Health Promotion, Chapter].

86 Global Economy and Development Program 5 in Promoting Sustainable Strategies to Improve Briefing Paper 02. iiG, [Accessed from www.iig. Access to Health Care in the Asian and Pacific Region. United Nations Economic and Social nya-cdf.pdf]. Commission for Asia and the Pacific (ESCAP), United Nations Publications. 23-164. [Accessed Sayed, Y. and Ahmed, R. (2008) Education decentrali- from sation in South Africa: Equity and participation in asp?id=1307]. the governance of schools. Paper Commissioned for the EFA Global Monitoring Report 2009, Quan, H.V. (2009) Programme 135-Sharing lessons Overcoming Inequality: why governance matters. on poverty reduction and development schemes for ethnic minorities in Vietnam. [Accessed from Schroeder, L. and Smoke, P. (2003) Intergovernmental Fiscal Transfers: Concepts, International Quan.pdf]. Practice, and Policy Issues in Intergovernmental Fiscal Transfers in Asia: Current Practices and Rao, G. and Khumalo, B. (2004) Sharing the Challenges for the Future. Edited by P. Smoke and Cake: A Review of Provincial Equitable Share YH. Kim. Asian Development Bank. Formula in South Africa. Paper presented at Intergovernmental Fiscal Relations Conference: Sen, A. (1992) Inequality Reexamined. Cambridge, MA: Ten Years of the Financial and Fiscal Commission Harvard University Press. (FCC)-Consolidation for Greater Equity. Cape Siringi, S. (2011) Kenya: KCPE-Private Schools Still Town, 10-12 August, 2004. on Top. AllAfrica, 28 December 2011. [Accessed Ravallion, M. (2005) A Poverty-Inequality Trade-Off? 4 January 2012, from: Journal of Economic Inequality August 2005. ries/201112290076.html]. 3(2). 169-182. Somerset, A. (2008) Access to Primary Education in Ravallion, M. (2009) A Comparative Perspective Kenya: A case study of the impact of the 2003 on Poverty Reduction in Brazil, China and FPE initiative at nine schools in Nairobi City India. Policy Research Working Paper 5080. and Nyeri District. CREATE Work in Progress. Washington DC: The World Bank. Brighton: University of Sussex.

Rokx, C., Schieber, G., Harimurti, P., Tandon, A., Steffensen, J. (2010) Fiscal Decentralisation and and Somanthan A. (2009) Health Financing in Sector Funding Principles and Practices-Annex Indonesia: A Reform Road Map. Washington D.C.: 4 Introduction to the Principles for Design The World Bank. of Intergovernmental Fiscal Transfer. Danida [Accessed from]. Romero, R. (2009) Decentralization, Accountability and the MPs Elections: The Case of the Sundet, G. and Moen, E. (2010) Political Economy Constituency Development Fund in Kenya. Analysis of Kenya. Norad Report 19/2009 Improving Institutions for Pro-Poor Growth (iiG) Discussion. Oslo: NORAD.

financing for a fairer , more prosperous kenya 87 The Economist (2001) Talking about a devolution-Indo- USAID (2012) Special Report: Livestock exports from nesia in The Economist 4 January 2001. Print edi- northern ports in the Horn of Africa, January 11, tion. 2012. Washington D.C.: USAID. [Accessed 10 September 2011]. Uwezo Kenya (2011) Are Our Children Learning?: Annual The Lancet (2007) Child Development in Developing Learning Assessment Report Nairobi: Uwezo Countries Series. 6 January 2007. The Lancet. Kenya. [Accessed from child-development-in-developing-countries]. Vandmoortele, J. (2009) Taking the MDGs beyond 2015: Hasten Slowly. Paper presented at High- The Lancet (2008) Maternal and Child Undernutrition Level Policy Forum: After 2015: Promoting Pro- Series. 16 January 2008. The Lancet. [Accessed Poor Policy after the MDGs, organized by DSA/ from EADI/Action Aid. Brussels, June 2009. and-child-undernutrition]. Wasanga, P., Ogle, M, and Wambua, R. (2010) The The Lancet (2011) Child Development in Developing Report on Monitoring Learner Achievement Study Countries Series 2. 23 September 2011. The for Class 3 in Literacy and Numeracy. Nairobi: The Lancet. [Accessed from http://www.thelancet. Kenya National Examinations Council. com/series/child-development-in-developing- countries-2]. West, A. (2009) Redistribution and Financing Schools in England under Labour: Are Resources Thuat, T. and Quan, H. (2008) A Case Study of the Going Where Needs Are Greatest? Educational Program for Socio-economic Development of Management Administration & Leadership 37(2). Communes Facing Extreme Difficulties in Ethnic 158-179. Minority and Mountainous Areas of Vietnam. Workshop on Strengthening the Development Wilkinson, R. and Pickett, K (2009) The Spirit Level: Results and Impacts of the Paris Declaration Why equality is better for everyone. London: Through Work on Gender Equality, Social Exclusion Penguin. and Human Rights. London, 12-13 March 2008. World Bank and the Asian Development Bank (2005) Uganda Multi-Donor Group (2007) Northern Uganda Decentralization in the Philippines: Strengthening Public Expenditure Review (NUPER), Final Local Government Financing & Resource Report January 2007. Hillsborough, UK: Regional Management in the Short Term. Washington D.C.: Forecasts Limited. The World Bank.

UNESCO (2010) Reaching the Marginalized. EFA Global World Bank (2005) Cross sectoral issues: Vo. 1 of Monitoring Report 2010. Paris: UNESCO. Vietnam-Managing public expenditure for poverty reduction and growth: public expenditure review UNESCO (2011) The Hidden Crisis: Armed Conflict and and integrated fiduciary assessment. Washington Education. EFA Global Monitoring Report 2011. D.C.: The World Bank. Paris: UNESCO.

88 Global Economy and Development Program World Bank (2006) World Development Report 2006: World Bank (2011b) Navigating the Storm, Delivering Equity and Development. Washington D.C.: The the promise: with a special focus on Kenya’s mo- World Bank. mentous devolution. Kenya Economic Update December 2011 Edition No. 5. Nairobi: The World World Bank (2009) Kenya Poverty and Inequality Bank. Assessment: Volume 1: Executive Summary and Synthesis Report. Report No.44190-KE. World Bank (2012) Tanzania Economic Update: Washington D.C.: The World Bank. Stairways to Heaven-Fiscal Prudence, Value for Money in Education, Economic Transformation World Bank (2010) Kenya Open Data 2009-10. [Accessed of Firms. February 2012, Issue 1. The World Bank. from ublic+Finance].

World Bank (2011a) The Drought and Food Crisis in the Horn of Africa: Impacts and Proposed Policy Responses for Kenya. Nairobi: The World Bank.

financing for a fairer , more prosperous kenya 89 ENDNOTES 5. The poverty gap is obtained by adding up all the shortfalls of the poor (considering the non-poor 1. The Arid and Semi-Arid Land counties as a group have a shortfall of zero) and dividing the total by represent 23 of the 47 counties created under the the population. Put differently, it gives the total devolved government system. In this report, we resources needed to bring all the poor to the level cover a sub-set of 12 primarily arid and pastoralist of the poverty line (divided by the number of indi- counties. viduals in the population). 2. The Local Authority Transfer Fund (LATF) is dis- 6. The North Eastern province was one of 8 prov- tributed on the basis of a formula that includes inces. The 2010 Constitution replaces the previ- (i) a basic minimum lump sum (ii) a 60 percent ous provincial structure with a new county-based weighting for population size and (iii) the relative structure. size of the urban and rural population of each lo- cal authority (with a weighting in favor of urban 7. The reported primary school completion rate for populations). [see (Government of Kenya, 2010) 2008 was 85 percent for boys and 75 percent for for the formula]. girls.

3. The income gap ratio is the difference between 8. The census reports a school-age population of 8.6 the poverty line and the average income (or million consumption) of the population living under the 9. One survey in 2008 found that the average costs poverty threshold expressed as a fraction of the of sending a child to secondary school fell from poverty line. around $185 to $79 (at prevailing exchange rates). 4. The 1997 survey reported a national poverty inci- dence of 52 percent.

90 Global Economy and Development Program The views expressed in this working paper do not necessarily reflect the official position of Brookings, its board or the advisory council members .

© 2012 The Brookings Institution

ISSN 2158-7779

Selected photos courtesy of the World Bank: cover left to right: Simone D . McCourtie (#1, #6), Masaru Goto (#2), Curt Carnemark (#3, #4, #5, #7) 1775 Massachusetts Avenue, NW Washington, DC 20036 202-797-6000